All patches and comments are welcome. Please squash your changes to logical
commits before using git-format-patch and git-send-email to
patches@git.madduck.net.
If you'd read over the Git project's submission guidelines and adhered to them,
I'd be especially grateful.
2 from asyncio.base_events import BaseEventLoop
3 from concurrent.futures import Executor, ProcessPoolExecutor
4 from datetime import datetime
5 from enum import Enum, Flag
6 from functools import lru_cache, partial, wraps
10 from multiprocessing import Manager
12 from pathlib import Path
37 from appdirs import user_cache_dir
38 from attr import dataclass, Factory
43 from blib2to3.pytree import Node, Leaf, type_repr
44 from blib2to3 import pygram, pytree
45 from blib2to3.pgen2 import driver, token
46 from blib2to3.pgen2.parse import ParseError
49 __version__ = "18.6b3"
50 DEFAULT_LINE_LENGTH = 88
52 r"/(\.git|\.hg|\.mypy_cache|\.tox|\.venv|_build|buck-out|build|dist)/"
54 DEFAULT_INCLUDES = r"\.pyi?$"
55 CACHE_DIR = Path(user_cache_dir("black", version=__version__))
67 LN = Union[Leaf, Node]
68 SplitFunc = Callable[["Line", bool], Iterator["Line"]]
71 CacheInfo = Tuple[Timestamp, FileSize]
72 Cache = Dict[Path, CacheInfo]
73 out = partial(click.secho, bold=True, err=True)
74 err = partial(click.secho, fg="red", err=True)
76 pygram.initialize(CACHE_DIR)
77 syms = pygram.python_symbols
80 class NothingChanged(UserWarning):
81 """Raised by :func:`format_file` when reformatted code is the same as source."""
84 class CannotSplit(Exception):
85 """A readable split that fits the allotted line length is impossible.
87 Raised by :func:`left_hand_split`, :func:`right_hand_split`, and
88 :func:`delimiter_split`.
92 class WriteBack(Enum):
98 def from_configuration(cls, *, check: bool, diff: bool) -> "WriteBack":
99 if check and not diff:
102 return cls.DIFF if diff else cls.YES
111 class FileMode(Flag):
115 NO_STRING_NORMALIZATION = 4
118 def from_configuration(
119 cls, *, py36: bool, pyi: bool, skip_string_normalization: bool
121 mode = cls.AUTO_DETECT
126 if skip_string_normalization:
127 mode |= cls.NO_STRING_NORMALIZATION
131 def read_pyproject_toml(
132 ctx: click.Context, param: click.Parameter, value: Union[str, int, bool, None]
134 """Inject Black configuration from "pyproject.toml" into defaults in `ctx`.
136 Returns the path to a successfully found and read configuration file, None
139 assert not isinstance(value, (int, bool)), "Invalid parameter type passed"
141 root = find_project_root(ctx.params.get("src", ()))
142 path = root / "pyproject.toml"
149 pyproject_toml = toml.load(value)
150 config = pyproject_toml.get("tool", {}).get("black", {})
151 except (toml.TomlDecodeError, OSError) as e:
152 raise click.BadOptionUsage(f"Error reading configuration file: {e}", ctx)
157 if ctx.default_map is None:
159 ctx.default_map.update( # type: ignore # bad types in .pyi
160 {k.replace("--", "").replace("-", "_"): v for k, v in config.items()}
165 @click.command(context_settings=dict(help_option_names=["-h", "--help"]))
170 default=DEFAULT_LINE_LENGTH,
171 help="How many character per line to allow.",
178 "Allow using Python 3.6-only syntax on all input files. This will put "
179 "trailing commas in function signatures and calls also after *args and "
180 "**kwargs. [default: per-file auto-detection]"
187 "Format all input files like typing stubs regardless of file extension "
188 "(useful when piping source on standard input)."
193 "--skip-string-normalization",
195 help="Don't normalize string quotes or prefixes.",
201 "Don't write the files back, just return the status. Return code 0 "
202 "means nothing would change. Return code 1 means some files would be "
203 "reformatted. Return code 123 means there was an internal error."
209 help="Don't write the files back, just output a diff for each file on stdout.",
214 help="If --fast given, skip temporary sanity checks. [default: --safe]",
219 default=DEFAULT_INCLUDES,
221 "A regular expression that matches files and directories that should be "
222 "included on recursive searches. An empty value means all files are "
223 "included regardless of the name. Use forward slashes for directories on "
224 "all platforms (Windows, too). Exclusions are calculated first, inclusions "
232 default=DEFAULT_EXCLUDES,
234 "A regular expression that matches files and directories that should be "
235 "excluded on recursive searches. An empty value means no paths are excluded. "
236 "Use forward slashes for directories on all platforms (Windows, too). "
237 "Exclusions are calculated first, inclusions later."
246 "Don't emit non-error messages to stderr. Errors are still emitted, "
247 "silence those with 2>/dev/null."
255 "Also emit messages to stderr about files that were not changed or were "
256 "ignored due to --exclude=."
259 @click.version_option(version=__version__)
264 exists=True, file_okay=True, dir_okay=True, readable=True, allow_dash=True
271 exists=False, file_okay=True, dir_okay=False, readable=True, allow_dash=False
274 callback=read_pyproject_toml,
275 help="Read configuration from PATH.",
286 skip_string_normalization: bool,
292 config: Optional[str],
294 """The uncompromising code formatter."""
295 write_back = WriteBack.from_configuration(check=check, diff=diff)
296 mode = FileMode.from_configuration(
297 py36=py36, pyi=pyi, skip_string_normalization=skip_string_normalization
299 if config and verbose:
300 out(f"Using configuration from {config}.", bold=False, fg="blue")
302 include_regex = re_compile_maybe_verbose(include)
304 err(f"Invalid regular expression for include given: {include!r}")
307 exclude_regex = re_compile_maybe_verbose(exclude)
309 err(f"Invalid regular expression for exclude given: {exclude!r}")
311 report = Report(check=check, quiet=quiet, verbose=verbose)
312 root = find_project_root(src)
313 sources: Set[Path] = set()
318 gen_python_files_in_dir(p, root, include_regex, exclude_regex, report)
320 elif p.is_file() or s == "-":
321 # if a file was explicitly given, we don't care about its extension
324 err(f"invalid path: {s}")
325 if len(sources) == 0:
326 if verbose or not quiet:
327 out("No paths given. Nothing to do 😴")
330 if len(sources) == 1:
333 line_length=line_length,
335 write_back=write_back,
340 loop = asyncio.get_event_loop()
341 executor = ProcessPoolExecutor(max_workers=os.cpu_count())
343 loop.run_until_complete(
346 line_length=line_length,
348 write_back=write_back,
357 if verbose or not quiet:
358 bang = "💥 💔 💥" if report.return_code else "✨ 🍰 ✨"
359 out(f"All done! {bang}")
360 click.secho(str(report), err=True)
361 ctx.exit(report.return_code)
368 write_back: WriteBack,
372 """Reformat a single file under `src` without spawning child processes.
374 If `quiet` is True, non-error messages are not output. `line_length`,
375 `write_back`, `fast` and `pyi` options are passed to
376 :func:`format_file_in_place` or :func:`format_stdin_to_stdout`.
380 if not src.is_file() and str(src) == "-":
381 if format_stdin_to_stdout(
382 line_length=line_length, fast=fast, write_back=write_back, mode=mode
384 changed = Changed.YES
387 if write_back != WriteBack.DIFF:
388 cache = read_cache(line_length, mode)
389 res_src = src.resolve()
390 if res_src in cache and cache[res_src] == get_cache_info(res_src):
391 changed = Changed.CACHED
392 if changed is not Changed.CACHED and format_file_in_place(
394 line_length=line_length,
396 write_back=write_back,
399 changed = Changed.YES
400 if write_back == WriteBack.YES and changed is not Changed.NO:
401 write_cache(cache, [src], line_length, mode)
402 report.done(src, changed)
403 except Exception as exc:
404 report.failed(src, str(exc))
407 async def schedule_formatting(
411 write_back: WriteBack,
417 """Run formatting of `sources` in parallel using the provided `executor`.
419 (Use ProcessPoolExecutors for actual parallelism.)
421 `line_length`, `write_back`, `fast`, and `pyi` options are passed to
422 :func:`format_file_in_place`.
425 if write_back != WriteBack.DIFF:
426 cache = read_cache(line_length, mode)
427 sources, cached = filter_cached(cache, sources)
428 for src in sorted(cached):
429 report.done(src, Changed.CACHED)
434 if write_back == WriteBack.DIFF:
435 # For diff output, we need locks to ensure we don't interleave output
436 # from different processes.
438 lock = manager.Lock()
440 loop.run_in_executor(
442 format_file_in_place,
450 for src in sorted(sources)
452 pending: Iterable[asyncio.Task] = tasks.keys()
454 loop.add_signal_handler(signal.SIGINT, cancel, pending)
455 loop.add_signal_handler(signal.SIGTERM, cancel, pending)
456 except NotImplementedError:
457 # There are no good alternatives for these on Windows
460 done, _ = await asyncio.wait(pending, return_when=asyncio.FIRST_COMPLETED)
462 src = tasks.pop(task)
464 cancelled.append(task)
465 elif task.exception():
466 report.failed(src, str(task.exception()))
468 formatted.append(src)
469 report.done(src, Changed.YES if task.result() else Changed.NO)
471 await asyncio.gather(*cancelled, loop=loop, return_exceptions=True)
472 if write_back == WriteBack.YES and formatted:
473 write_cache(cache, formatted, line_length, mode)
476 def format_file_in_place(
480 write_back: WriteBack = WriteBack.NO,
481 mode: FileMode = FileMode.AUTO_DETECT,
482 lock: Any = None, # multiprocessing.Manager().Lock() is some crazy proxy
484 """Format file under `src` path. Return True if changed.
486 If `write_back` is True, write reformatted code back to stdout.
487 `line_length` and `fast` options are passed to :func:`format_file_contents`.
489 if src.suffix == ".pyi":
492 then = datetime.utcfromtimestamp(src.stat().st_mtime)
493 with open(src, "rb") as buf:
494 src_contents, encoding, newline = decode_bytes(buf.read())
496 dst_contents = format_file_contents(
497 src_contents, line_length=line_length, fast=fast, mode=mode
499 except NothingChanged:
502 if write_back == write_back.YES:
503 with open(src, "w", encoding=encoding, newline=newline) as f:
504 f.write(dst_contents)
505 elif write_back == write_back.DIFF:
506 now = datetime.utcnow()
507 src_name = f"{src}\t{then} +0000"
508 dst_name = f"{src}\t{now} +0000"
509 diff_contents = diff(src_contents, dst_contents, src_name, dst_name)
513 f = io.TextIOWrapper(
519 f.write(diff_contents)
527 def format_stdin_to_stdout(
530 write_back: WriteBack = WriteBack.NO,
531 mode: FileMode = FileMode.AUTO_DETECT,
533 """Format file on stdin. Return True if changed.
535 If `write_back` is True, write reformatted code back to stdout.
536 `line_length`, `fast`, `is_pyi`, and `force_py36` arguments are passed to
537 :func:`format_file_contents`.
539 then = datetime.utcnow()
540 src, encoding, newline = decode_bytes(sys.stdin.buffer.read())
543 dst = format_file_contents(src, line_length=line_length, fast=fast, mode=mode)
546 except NothingChanged:
550 f = io.TextIOWrapper(
551 sys.stdout.buffer, encoding=encoding, newline=newline, write_through=True
553 if write_back == WriteBack.YES:
555 elif write_back == WriteBack.DIFF:
556 now = datetime.utcnow()
557 src_name = f"STDIN\t{then} +0000"
558 dst_name = f"STDOUT\t{now} +0000"
559 f.write(diff(src, dst, src_name, dst_name))
563 def format_file_contents(
568 mode: FileMode = FileMode.AUTO_DETECT,
570 """Reformat contents a file and return new contents.
572 If `fast` is False, additionally confirm that the reformatted code is
573 valid by calling :func:`assert_equivalent` and :func:`assert_stable` on it.
574 `line_length` is passed to :func:`format_str`.
576 if src_contents.strip() == "":
579 dst_contents = format_str(src_contents, line_length=line_length, mode=mode)
580 if src_contents == dst_contents:
584 assert_equivalent(src_contents, dst_contents)
585 assert_stable(src_contents, dst_contents, line_length=line_length, mode=mode)
590 src_contents: str, line_length: int, *, mode: FileMode = FileMode.AUTO_DETECT
592 """Reformat a string and return new contents.
594 `line_length` determines how many characters per line are allowed.
596 src_node = lib2to3_parse(src_contents)
598 future_imports = get_future_imports(src_node)
599 is_pyi = bool(mode & FileMode.PYI)
600 py36 = bool(mode & FileMode.PYTHON36) or is_python36(src_node)
601 normalize_strings = not bool(mode & FileMode.NO_STRING_NORMALIZATION)
602 normalize_fmt_off(src_node)
603 lines = LineGenerator(
604 remove_u_prefix=py36 or "unicode_literals" in future_imports,
606 normalize_strings=normalize_strings,
608 elt = EmptyLineTracker(is_pyi=is_pyi)
611 for current_line in lines.visit(src_node):
612 for _ in range(after):
613 dst_contents += str(empty_line)
614 before, after = elt.maybe_empty_lines(current_line)
615 for _ in range(before):
616 dst_contents += str(empty_line)
617 for line in split_line(current_line, line_length=line_length, py36=py36):
618 dst_contents += str(line)
622 def decode_bytes(src: bytes) -> Tuple[FileContent, Encoding, NewLine]:
623 """Return a tuple of (decoded_contents, encoding, newline).
625 `newline` is either CRLF or LF but `decoded_contents` is decoded with
626 universal newlines (i.e. only contains LF).
628 srcbuf = io.BytesIO(src)
629 encoding, lines = tokenize.detect_encoding(srcbuf.readline)
631 return "", encoding, "\n"
633 newline = "\r\n" if b"\r\n" == lines[0][-2:] else "\n"
635 with io.TextIOWrapper(srcbuf, encoding) as tiow:
636 return tiow.read(), encoding, newline
640 pygram.python_grammar_no_print_statement_no_exec_statement,
641 pygram.python_grammar_no_print_statement,
642 pygram.python_grammar,
646 def lib2to3_parse(src_txt: str) -> Node:
647 """Given a string with source, return the lib2to3 Node."""
648 grammar = pygram.python_grammar_no_print_statement
649 if src_txt[-1:] != "\n":
651 for grammar in GRAMMARS:
652 drv = driver.Driver(grammar, pytree.convert)
654 result = drv.parse_string(src_txt, True)
657 except ParseError as pe:
658 lineno, column = pe.context[1]
659 lines = src_txt.splitlines()
661 faulty_line = lines[lineno - 1]
663 faulty_line = "<line number missing in source>"
664 exc = ValueError(f"Cannot parse: {lineno}:{column}: {faulty_line}")
668 if isinstance(result, Leaf):
669 result = Node(syms.file_input, [result])
673 def lib2to3_unparse(node: Node) -> str:
674 """Given a lib2to3 node, return its string representation."""
682 class Visitor(Generic[T]):
683 """Basic lib2to3 visitor that yields things of type `T` on `visit()`."""
685 def visit(self, node: LN) -> Iterator[T]:
686 """Main method to visit `node` and its children.
688 It tries to find a `visit_*()` method for the given `node.type`, like
689 `visit_simple_stmt` for Node objects or `visit_INDENT` for Leaf objects.
690 If no dedicated `visit_*()` method is found, chooses `visit_default()`
693 Then yields objects of type `T` from the selected visitor.
696 name = token.tok_name[node.type]
698 name = type_repr(node.type)
699 yield from getattr(self, f"visit_{name}", self.visit_default)(node)
701 def visit_default(self, node: LN) -> Iterator[T]:
702 """Default `visit_*()` implementation. Recurses to children of `node`."""
703 if isinstance(node, Node):
704 for child in node.children:
705 yield from self.visit(child)
709 class DebugVisitor(Visitor[T]):
712 def visit_default(self, node: LN) -> Iterator[T]:
713 indent = " " * (2 * self.tree_depth)
714 if isinstance(node, Node):
715 _type = type_repr(node.type)
716 out(f"{indent}{_type}", fg="yellow")
718 for child in node.children:
719 yield from self.visit(child)
722 out(f"{indent}/{_type}", fg="yellow", bold=False)
724 _type = token.tok_name.get(node.type, str(node.type))
725 out(f"{indent}{_type}", fg="blue", nl=False)
727 # We don't have to handle prefixes for `Node` objects since
728 # that delegates to the first child anyway.
729 out(f" {node.prefix!r}", fg="green", bold=False, nl=False)
730 out(f" {node.value!r}", fg="blue", bold=False)
733 def show(cls, code: Union[str, Leaf, Node]) -> None:
734 """Pretty-print the lib2to3 AST of a given string of `code`.
736 Convenience method for debugging.
738 v: DebugVisitor[None] = DebugVisitor()
739 if isinstance(code, str):
740 code = lib2to3_parse(code)
744 KEYWORDS = set(keyword.kwlist)
745 WHITESPACE = {token.DEDENT, token.INDENT, token.NEWLINE}
746 FLOW_CONTROL = {"return", "raise", "break", "continue"}
757 STANDALONE_COMMENT = 153
758 token.tok_name[STANDALONE_COMMENT] = "STANDALONE_COMMENT"
759 LOGIC_OPERATORS = {"and", "or"}
784 STARS = {token.STAR, token.DOUBLESTAR}
787 syms.argument, # double star in arglist
788 syms.trailer, # single argument to call
790 syms.varargslist, # lambdas
792 UNPACKING_PARENTS = {
793 syms.atom, # single element of a list or set literal
797 syms.testlist_star_expr,
799 SURROUNDED_BY_BRACKETS = {
804 syms.import_as_names,
807 syms.testlist_star_expr,
844 COMPREHENSION_PRIORITY = 20
846 TERNARY_PRIORITY = 16
849 COMPARATOR_PRIORITY = 10
860 token.DOUBLESLASH: 4,
870 class BracketTracker:
871 """Keeps track of brackets on a line."""
874 bracket_match: Dict[Tuple[Depth, NodeType], Leaf] = Factory(dict)
875 delimiters: Dict[LeafID, Priority] = Factory(dict)
876 previous: Optional[Leaf] = None
877 _for_loop_variable: int = 0
878 _lambda_arguments: int = 0
880 def mark(self, leaf: Leaf) -> None:
881 """Mark `leaf` with bracket-related metadata. Keep track of delimiters.
883 All leaves receive an int `bracket_depth` field that stores how deep
884 within brackets a given leaf is. 0 means there are no enclosing brackets
885 that started on this line.
887 If a leaf is itself a closing bracket, it receives an `opening_bracket`
888 field that it forms a pair with. This is a one-directional link to
889 avoid reference cycles.
891 If a leaf is a delimiter (a token on which Black can split the line if
892 needed) and it's on depth 0, its `id()` is stored in the tracker's
895 if leaf.type == token.COMMENT:
898 self.maybe_decrement_after_for_loop_variable(leaf)
899 self.maybe_decrement_after_lambda_arguments(leaf)
900 if leaf.type in CLOSING_BRACKETS:
902 opening_bracket = self.bracket_match.pop((self.depth, leaf.type))
903 leaf.opening_bracket = opening_bracket
904 leaf.bracket_depth = self.depth
906 delim = is_split_before_delimiter(leaf, self.previous)
907 if delim and self.previous is not None:
908 self.delimiters[id(self.previous)] = delim
910 delim = is_split_after_delimiter(leaf, self.previous)
912 self.delimiters[id(leaf)] = delim
913 if leaf.type in OPENING_BRACKETS:
914 self.bracket_match[self.depth, BRACKET[leaf.type]] = leaf
917 self.maybe_increment_lambda_arguments(leaf)
918 self.maybe_increment_for_loop_variable(leaf)
920 def any_open_brackets(self) -> bool:
921 """Return True if there is an yet unmatched open bracket on the line."""
922 return bool(self.bracket_match)
924 def max_delimiter_priority(self, exclude: Iterable[LeafID] = ()) -> int:
925 """Return the highest priority of a delimiter found on the line.
927 Values are consistent with what `is_split_*_delimiter()` return.
928 Raises ValueError on no delimiters.
930 return max(v for k, v in self.delimiters.items() if k not in exclude)
932 def delimiter_count_with_priority(self, priority: int = 0) -> int:
933 """Return the number of delimiters with the given `priority`.
935 If no `priority` is passed, defaults to max priority on the line.
937 if not self.delimiters:
940 priority = priority or self.max_delimiter_priority()
941 return sum(1 for p in self.delimiters.values() if p == priority)
943 def maybe_increment_for_loop_variable(self, leaf: Leaf) -> bool:
944 """In a for loop, or comprehension, the variables are often unpacks.
946 To avoid splitting on the comma in this situation, increase the depth of
947 tokens between `for` and `in`.
949 if leaf.type == token.NAME and leaf.value == "for":
951 self._for_loop_variable += 1
956 def maybe_decrement_after_for_loop_variable(self, leaf: Leaf) -> bool:
957 """See `maybe_increment_for_loop_variable` above for explanation."""
958 if self._for_loop_variable and leaf.type == token.NAME and leaf.value == "in":
960 self._for_loop_variable -= 1
965 def maybe_increment_lambda_arguments(self, leaf: Leaf) -> bool:
966 """In a lambda expression, there might be more than one argument.
968 To avoid splitting on the comma in this situation, increase the depth of
969 tokens between `lambda` and `:`.
971 if leaf.type == token.NAME and leaf.value == "lambda":
973 self._lambda_arguments += 1
978 def maybe_decrement_after_lambda_arguments(self, leaf: Leaf) -> bool:
979 """See `maybe_increment_lambda_arguments` above for explanation."""
980 if self._lambda_arguments and leaf.type == token.COLON:
982 self._lambda_arguments -= 1
987 def get_open_lsqb(self) -> Optional[Leaf]:
988 """Return the most recent opening square bracket (if any)."""
989 return self.bracket_match.get((self.depth - 1, token.RSQB))
994 """Holds leaves and comments. Can be printed with `str(line)`."""
997 leaves: List[Leaf] = Factory(list)
998 comments: List[Tuple[Index, Leaf]] = Factory(list)
999 bracket_tracker: BracketTracker = Factory(BracketTracker)
1000 inside_brackets: bool = False
1001 should_explode: bool = False
1003 def append(self, leaf: Leaf, preformatted: bool = False) -> None:
1004 """Add a new `leaf` to the end of the line.
1006 Unless `preformatted` is True, the `leaf` will receive a new consistent
1007 whitespace prefix and metadata applied by :class:`BracketTracker`.
1008 Trailing commas are maybe removed, unpacked for loop variables are
1009 demoted from being delimiters.
1011 Inline comments are put aside.
1013 has_value = leaf.type in BRACKETS or bool(leaf.value.strip())
1017 if token.COLON == leaf.type and self.is_class_paren_empty:
1018 del self.leaves[-2:]
1019 if self.leaves and not preformatted:
1020 # Note: at this point leaf.prefix should be empty except for
1021 # imports, for which we only preserve newlines.
1022 leaf.prefix += whitespace(
1023 leaf, complex_subscript=self.is_complex_subscript(leaf)
1025 if self.inside_brackets or not preformatted:
1026 self.bracket_tracker.mark(leaf)
1027 self.maybe_remove_trailing_comma(leaf)
1028 if not self.append_comment(leaf):
1029 self.leaves.append(leaf)
1031 def append_safe(self, leaf: Leaf, preformatted: bool = False) -> None:
1032 """Like :func:`append()` but disallow invalid standalone comment structure.
1034 Raises ValueError when any `leaf` is appended after a standalone comment
1035 or when a standalone comment is not the first leaf on the line.
1037 if self.bracket_tracker.depth == 0:
1039 raise ValueError("cannot append to standalone comments")
1041 if self.leaves and leaf.type == STANDALONE_COMMENT:
1043 "cannot append standalone comments to a populated line"
1046 self.append(leaf, preformatted=preformatted)
1049 def is_comment(self) -> bool:
1050 """Is this line a standalone comment?"""
1051 return len(self.leaves) == 1 and self.leaves[0].type == STANDALONE_COMMENT
1054 def is_decorator(self) -> bool:
1055 """Is this line a decorator?"""
1056 return bool(self) and self.leaves[0].type == token.AT
1059 def is_import(self) -> bool:
1060 """Is this an import line?"""
1061 return bool(self) and is_import(self.leaves[0])
1064 def is_class(self) -> bool:
1065 """Is this line a class definition?"""
1068 and self.leaves[0].type == token.NAME
1069 and self.leaves[0].value == "class"
1073 def is_stub_class(self) -> bool:
1074 """Is this line a class definition with a body consisting only of "..."?"""
1075 return self.is_class and self.leaves[-3:] == [
1076 Leaf(token.DOT, ".") for _ in range(3)
1080 def is_def(self) -> bool:
1081 """Is this a function definition? (Also returns True for async defs.)"""
1083 first_leaf = self.leaves[0]
1088 second_leaf: Optional[Leaf] = self.leaves[1]
1091 return (first_leaf.type == token.NAME and first_leaf.value == "def") or (
1092 first_leaf.type == token.ASYNC
1093 and second_leaf is not None
1094 and second_leaf.type == token.NAME
1095 and second_leaf.value == "def"
1099 def is_class_paren_empty(self) -> bool:
1100 """Is this a class with no base classes but using parentheses?
1102 Those are unnecessary and should be removed.
1106 and len(self.leaves) == 4
1108 and self.leaves[2].type == token.LPAR
1109 and self.leaves[2].value == "("
1110 and self.leaves[3].type == token.RPAR
1111 and self.leaves[3].value == ")"
1115 def is_triple_quoted_string(self) -> bool:
1116 """Is the line a triple quoted string?"""
1119 and self.leaves[0].type == token.STRING
1120 and self.leaves[0].value.startswith(('"""', "'''"))
1123 def contains_standalone_comments(self, depth_limit: int = sys.maxsize) -> bool:
1124 """If so, needs to be split before emitting."""
1125 for leaf in self.leaves:
1126 if leaf.type == STANDALONE_COMMENT:
1127 if leaf.bracket_depth <= depth_limit:
1132 def contains_multiline_strings(self) -> bool:
1133 for leaf in self.leaves:
1134 if is_multiline_string(leaf):
1139 def maybe_remove_trailing_comma(self, closing: Leaf) -> bool:
1140 """Remove trailing comma if there is one and it's safe."""
1143 and self.leaves[-1].type == token.COMMA
1144 and closing.type in CLOSING_BRACKETS
1148 if closing.type == token.RBRACE:
1149 self.remove_trailing_comma()
1152 if closing.type == token.RSQB:
1153 comma = self.leaves[-1]
1154 if comma.parent and comma.parent.type == syms.listmaker:
1155 self.remove_trailing_comma()
1158 # For parens let's check if it's safe to remove the comma.
1159 # Imports are always safe.
1161 self.remove_trailing_comma()
1164 # Otheriwsse, if the trailing one is the only one, we might mistakenly
1165 # change a tuple into a different type by removing the comma.
1166 depth = closing.bracket_depth + 1
1168 opening = closing.opening_bracket
1169 for _opening_index, leaf in enumerate(self.leaves):
1176 for leaf in self.leaves[_opening_index + 1 :]:
1180 bracket_depth = leaf.bracket_depth
1181 if bracket_depth == depth and leaf.type == token.COMMA:
1183 if leaf.parent and leaf.parent.type == syms.arglist:
1188 self.remove_trailing_comma()
1193 def append_comment(self, comment: Leaf) -> bool:
1194 """Add an inline or standalone comment to the line."""
1196 comment.type == STANDALONE_COMMENT
1197 and self.bracket_tracker.any_open_brackets()
1202 if comment.type != token.COMMENT:
1205 after = len(self.leaves) - 1
1207 comment.type = STANDALONE_COMMENT
1212 self.comments.append((after, comment))
1215 def comments_after(self, leaf: Leaf, _index: int = -1) -> Iterator[Leaf]:
1216 """Generate comments that should appear directly after `leaf`.
1218 Provide a non-negative leaf `_index` to speed up the function.
1220 if not self.comments:
1224 for _index, _leaf in enumerate(self.leaves):
1231 for index, comment_after in self.comments:
1235 def remove_trailing_comma(self) -> None:
1236 """Remove the trailing comma and moves the comments attached to it."""
1237 comma_index = len(self.leaves) - 1
1238 for i in range(len(self.comments)):
1239 comment_index, comment = self.comments[i]
1240 if comment_index == comma_index:
1241 self.comments[i] = (comma_index - 1, comment)
1244 def is_complex_subscript(self, leaf: Leaf) -> bool:
1245 """Return True iff `leaf` is part of a slice with non-trivial exprs."""
1246 open_lsqb = self.bracket_tracker.get_open_lsqb()
1247 if open_lsqb is None:
1250 subscript_start = open_lsqb.next_sibling
1252 if isinstance(subscript_start, Node):
1253 if subscript_start.type == syms.listmaker:
1256 if subscript_start.type == syms.subscriptlist:
1257 subscript_start = child_towards(subscript_start, leaf)
1258 return subscript_start is not None and any(
1259 n.type in TEST_DESCENDANTS for n in subscript_start.pre_order()
1262 def __str__(self) -> str:
1263 """Render the line."""
1267 indent = " " * self.depth
1268 leaves = iter(self.leaves)
1269 first = next(leaves)
1270 res = f"{first.prefix}{indent}{first.value}"
1273 for _, comment in self.comments:
1277 def __bool__(self) -> bool:
1278 """Return True if the line has leaves or comments."""
1279 return bool(self.leaves or self.comments)
1283 class EmptyLineTracker:
1284 """Provides a stateful method that returns the number of potential extra
1285 empty lines needed before and after the currently processed line.
1287 Note: this tracker works on lines that haven't been split yet. It assumes
1288 the prefix of the first leaf consists of optional newlines. Those newlines
1289 are consumed by `maybe_empty_lines()` and included in the computation.
1292 is_pyi: bool = False
1293 previous_line: Optional[Line] = None
1294 previous_after: int = 0
1295 previous_defs: List[int] = Factory(list)
1297 def maybe_empty_lines(self, current_line: Line) -> Tuple[int, int]:
1298 """Return the number of extra empty lines before and after the `current_line`.
1300 This is for separating `def`, `async def` and `class` with extra empty
1301 lines (two on module-level).
1303 before, after = self._maybe_empty_lines(current_line)
1304 before -= self.previous_after
1305 self.previous_after = after
1306 self.previous_line = current_line
1307 return before, after
1309 def _maybe_empty_lines(self, current_line: Line) -> Tuple[int, int]:
1311 if current_line.depth == 0:
1312 max_allowed = 1 if self.is_pyi else 2
1313 if current_line.leaves:
1314 # Consume the first leaf's extra newlines.
1315 first_leaf = current_line.leaves[0]
1316 before = first_leaf.prefix.count("\n")
1317 before = min(before, max_allowed)
1318 first_leaf.prefix = ""
1321 depth = current_line.depth
1322 while self.previous_defs and self.previous_defs[-1] >= depth:
1323 self.previous_defs.pop()
1325 before = 0 if depth else 1
1327 before = 1 if depth else 2
1328 if current_line.is_decorator or current_line.is_def or current_line.is_class:
1329 return self._maybe_empty_lines_for_class_or_def(current_line, before)
1333 and self.previous_line.is_import
1334 and not current_line.is_import
1335 and depth == self.previous_line.depth
1337 return (before or 1), 0
1341 and self.previous_line.is_class
1342 and current_line.is_triple_quoted_string
1348 def _maybe_empty_lines_for_class_or_def(
1349 self, current_line: Line, before: int
1350 ) -> Tuple[int, int]:
1351 if not current_line.is_decorator:
1352 self.previous_defs.append(current_line.depth)
1353 if self.previous_line is None:
1354 # Don't insert empty lines before the first line in the file.
1357 if self.previous_line.is_decorator:
1360 if self.previous_line.depth < current_line.depth and (
1361 self.previous_line.is_class or self.previous_line.is_def
1366 self.previous_line.is_comment
1367 and self.previous_line.depth == current_line.depth
1373 if self.previous_line.depth > current_line.depth:
1375 elif current_line.is_class or self.previous_line.is_class:
1376 if current_line.is_stub_class and self.previous_line.is_stub_class:
1377 # No blank line between classes with an emty body
1381 elif current_line.is_def and not self.previous_line.is_def:
1382 # Blank line between a block of functions and a block of non-functions
1388 if current_line.depth and newlines:
1394 class LineGenerator(Visitor[Line]):
1395 """Generates reformatted Line objects. Empty lines are not emitted.
1397 Note: destroys the tree it's visiting by mutating prefixes of its leaves
1398 in ways that will no longer stringify to valid Python code on the tree.
1401 is_pyi: bool = False
1402 normalize_strings: bool = True
1403 current_line: Line = Factory(Line)
1404 remove_u_prefix: bool = False
1406 def line(self, indent: int = 0) -> Iterator[Line]:
1409 If the line is empty, only emit if it makes sense.
1410 If the line is too long, split it first and then generate.
1412 If any lines were generated, set up a new current_line.
1414 if not self.current_line:
1415 self.current_line.depth += indent
1416 return # Line is empty, don't emit. Creating a new one unnecessary.
1418 complete_line = self.current_line
1419 self.current_line = Line(depth=complete_line.depth + indent)
1422 def visit_default(self, node: LN) -> Iterator[Line]:
1423 """Default `visit_*()` implementation. Recurses to children of `node`."""
1424 if isinstance(node, Leaf):
1425 any_open_brackets = self.current_line.bracket_tracker.any_open_brackets()
1426 for comment in generate_comments(node):
1427 if any_open_brackets:
1428 # any comment within brackets is subject to splitting
1429 self.current_line.append(comment)
1430 elif comment.type == token.COMMENT:
1431 # regular trailing comment
1432 self.current_line.append(comment)
1433 yield from self.line()
1436 # regular standalone comment
1437 yield from self.line()
1439 self.current_line.append(comment)
1440 yield from self.line()
1442 normalize_prefix(node, inside_brackets=any_open_brackets)
1443 if self.normalize_strings and node.type == token.STRING:
1444 normalize_string_prefix(node, remove_u_prefix=self.remove_u_prefix)
1445 normalize_string_quotes(node)
1446 if node.type not in WHITESPACE:
1447 self.current_line.append(node)
1448 yield from super().visit_default(node)
1450 def visit_INDENT(self, node: Node) -> Iterator[Line]:
1451 """Increase indentation level, maybe yield a line."""
1452 # In blib2to3 INDENT never holds comments.
1453 yield from self.line(+1)
1454 yield from self.visit_default(node)
1456 def visit_DEDENT(self, node: Node) -> Iterator[Line]:
1457 """Decrease indentation level, maybe yield a line."""
1458 # The current line might still wait for trailing comments. At DEDENT time
1459 # there won't be any (they would be prefixes on the preceding NEWLINE).
1460 # Emit the line then.
1461 yield from self.line()
1463 # While DEDENT has no value, its prefix may contain standalone comments
1464 # that belong to the current indentation level. Get 'em.
1465 yield from self.visit_default(node)
1467 # Finally, emit the dedent.
1468 yield from self.line(-1)
1471 self, node: Node, keywords: Set[str], parens: Set[str]
1472 ) -> Iterator[Line]:
1473 """Visit a statement.
1475 This implementation is shared for `if`, `while`, `for`, `try`, `except`,
1476 `def`, `with`, `class`, `assert` and assignments.
1478 The relevant Python language `keywords` for a given statement will be
1479 NAME leaves within it. This methods puts those on a separate line.
1481 `parens` holds a set of string leaf values immediately after which
1482 invisible parens should be put.
1484 normalize_invisible_parens(node, parens_after=parens)
1485 for child in node.children:
1486 if child.type == token.NAME and child.value in keywords: # type: ignore
1487 yield from self.line()
1489 yield from self.visit(child)
1491 def visit_suite(self, node: Node) -> Iterator[Line]:
1492 """Visit a suite."""
1493 if self.is_pyi and is_stub_suite(node):
1494 yield from self.visit(node.children[2])
1496 yield from self.visit_default(node)
1498 def visit_simple_stmt(self, node: Node) -> Iterator[Line]:
1499 """Visit a statement without nested statements."""
1500 is_suite_like = node.parent and node.parent.type in STATEMENT
1502 if self.is_pyi and is_stub_body(node):
1503 yield from self.visit_default(node)
1505 yield from self.line(+1)
1506 yield from self.visit_default(node)
1507 yield from self.line(-1)
1510 if not self.is_pyi or not node.parent or not is_stub_suite(node.parent):
1511 yield from self.line()
1512 yield from self.visit_default(node)
1514 def visit_async_stmt(self, node: Node) -> Iterator[Line]:
1515 """Visit `async def`, `async for`, `async with`."""
1516 yield from self.line()
1518 children = iter(node.children)
1519 for child in children:
1520 yield from self.visit(child)
1522 if child.type == token.ASYNC:
1525 internal_stmt = next(children)
1526 for child in internal_stmt.children:
1527 yield from self.visit(child)
1529 def visit_decorators(self, node: Node) -> Iterator[Line]:
1530 """Visit decorators."""
1531 for child in node.children:
1532 yield from self.line()
1533 yield from self.visit(child)
1535 def visit_SEMI(self, leaf: Leaf) -> Iterator[Line]:
1536 """Remove a semicolon and put the other statement on a separate line."""
1537 yield from self.line()
1539 def visit_ENDMARKER(self, leaf: Leaf) -> Iterator[Line]:
1540 """End of file. Process outstanding comments and end with a newline."""
1541 yield from self.visit_default(leaf)
1542 yield from self.line()
1544 def visit_STANDALONE_COMMENT(self, leaf: Leaf) -> Iterator[Line]:
1545 if not self.current_line.bracket_tracker.any_open_brackets():
1546 yield from self.line()
1547 yield from self.visit_default(leaf)
1549 def __attrs_post_init__(self) -> None:
1550 """You are in a twisty little maze of passages."""
1553 self.visit_assert_stmt = partial(v, keywords={"assert"}, parens={"assert", ","})
1554 self.visit_if_stmt = partial(
1555 v, keywords={"if", "else", "elif"}, parens={"if", "elif"}
1557 self.visit_while_stmt = partial(v, keywords={"while", "else"}, parens={"while"})
1558 self.visit_for_stmt = partial(v, keywords={"for", "else"}, parens={"for", "in"})
1559 self.visit_try_stmt = partial(
1560 v, keywords={"try", "except", "else", "finally"}, parens=Ø
1562 self.visit_except_clause = partial(v, keywords={"except"}, parens=Ø)
1563 self.visit_with_stmt = partial(v, keywords={"with"}, parens=Ø)
1564 self.visit_funcdef = partial(v, keywords={"def"}, parens=Ø)
1565 self.visit_classdef = partial(v, keywords={"class"}, parens=Ø)
1566 self.visit_expr_stmt = partial(v, keywords=Ø, parens=ASSIGNMENTS)
1567 self.visit_return_stmt = partial(v, keywords={"return"}, parens={"return"})
1568 self.visit_import_from = partial(v, keywords=Ø, parens={"import"})
1569 self.visit_async_funcdef = self.visit_async_stmt
1570 self.visit_decorated = self.visit_decorators
1573 IMPLICIT_TUPLE = {syms.testlist, syms.testlist_star_expr, syms.exprlist}
1574 BRACKET = {token.LPAR: token.RPAR, token.LSQB: token.RSQB, token.LBRACE: token.RBRACE}
1575 OPENING_BRACKETS = set(BRACKET.keys())
1576 CLOSING_BRACKETS = set(BRACKET.values())
1577 BRACKETS = OPENING_BRACKETS | CLOSING_BRACKETS
1578 ALWAYS_NO_SPACE = CLOSING_BRACKETS | {token.COMMA, STANDALONE_COMMENT}
1581 def whitespace(leaf: Leaf, *, complex_subscript: bool) -> str: # noqa C901
1582 """Return whitespace prefix if needed for the given `leaf`.
1584 `complex_subscript` signals whether the given leaf is part of a subscription
1585 which has non-trivial arguments, like arithmetic expressions or function calls.
1593 if t in ALWAYS_NO_SPACE:
1596 if t == token.COMMENT:
1599 assert p is not None, f"INTERNAL ERROR: hand-made leaf without parent: {leaf!r}"
1600 if t == token.COLON and p.type not in {
1607 prev = leaf.prev_sibling
1609 prevp = preceding_leaf(p)
1610 if not prevp or prevp.type in OPENING_BRACKETS:
1613 if t == token.COLON:
1614 if prevp.type == token.COLON:
1617 elif prevp.type != token.COMMA and not complex_subscript:
1622 if prevp.type == token.EQUAL:
1624 if prevp.parent.type in {
1632 elif prevp.parent.type == syms.typedargslist:
1633 # A bit hacky: if the equal sign has whitespace, it means we
1634 # previously found it's a typed argument. So, we're using
1638 elif prevp.type in STARS:
1639 if is_vararg(prevp, within=VARARGS_PARENTS | UNPACKING_PARENTS):
1642 elif prevp.type == token.COLON:
1643 if prevp.parent and prevp.parent.type in {syms.subscript, syms.sliceop}:
1644 return SPACE if complex_subscript else NO
1648 and prevp.parent.type == syms.factor
1649 and prevp.type in MATH_OPERATORS
1654 prevp.type == token.RIGHTSHIFT
1656 and prevp.parent.type == syms.shift_expr
1657 and prevp.prev_sibling
1658 and prevp.prev_sibling.type == token.NAME
1659 and prevp.prev_sibling.value == "print" # type: ignore
1661 # Python 2 print chevron
1664 elif prev.type in OPENING_BRACKETS:
1667 if p.type in {syms.parameters, syms.arglist}:
1668 # untyped function signatures or calls
1669 if not prev or prev.type != token.COMMA:
1672 elif p.type == syms.varargslist:
1674 if prev and prev.type != token.COMMA:
1677 elif p.type == syms.typedargslist:
1678 # typed function signatures
1682 if t == token.EQUAL:
1683 if prev.type != syms.tname:
1686 elif prev.type == token.EQUAL:
1687 # A bit hacky: if the equal sign has whitespace, it means we
1688 # previously found it's a typed argument. So, we're using that, too.
1691 elif prev.type != token.COMMA:
1694 elif p.type == syms.tname:
1697 prevp = preceding_leaf(p)
1698 if not prevp or prevp.type != token.COMMA:
1701 elif p.type == syms.trailer:
1702 # attributes and calls
1703 if t == token.LPAR or t == token.RPAR:
1708 prevp = preceding_leaf(p)
1709 if not prevp or prevp.type != token.NUMBER:
1712 elif t == token.LSQB:
1715 elif prev.type != token.COMMA:
1718 elif p.type == syms.argument:
1720 if t == token.EQUAL:
1724 prevp = preceding_leaf(p)
1725 if not prevp or prevp.type == token.LPAR:
1728 elif prev.type in {token.EQUAL} | STARS:
1731 elif p.type == syms.decorator:
1735 elif p.type == syms.dotted_name:
1739 prevp = preceding_leaf(p)
1740 if not prevp or prevp.type == token.AT or prevp.type == token.DOT:
1743 elif p.type == syms.classdef:
1747 if prev and prev.type == token.LPAR:
1750 elif p.type in {syms.subscript, syms.sliceop}:
1753 assert p.parent is not None, "subscripts are always parented"
1754 if p.parent.type == syms.subscriptlist:
1759 elif not complex_subscript:
1762 elif p.type == syms.atom:
1763 if prev and t == token.DOT:
1764 # dots, but not the first one.
1767 elif p.type == syms.dictsetmaker:
1769 if prev and prev.type == token.DOUBLESTAR:
1772 elif p.type in {syms.factor, syms.star_expr}:
1775 prevp = preceding_leaf(p)
1776 if not prevp or prevp.type in OPENING_BRACKETS:
1779 prevp_parent = prevp.parent
1780 assert prevp_parent is not None
1781 if prevp.type == token.COLON and prevp_parent.type in {
1787 elif prevp.type == token.EQUAL and prevp_parent.type == syms.argument:
1790 elif t in {token.NAME, token.NUMBER, token.STRING}:
1793 elif p.type == syms.import_from:
1795 if prev and prev.type == token.DOT:
1798 elif t == token.NAME:
1802 if prev and prev.type == token.DOT:
1805 elif p.type == syms.sliceop:
1811 def preceding_leaf(node: Optional[LN]) -> Optional[Leaf]:
1812 """Return the first leaf that precedes `node`, if any."""
1814 res = node.prev_sibling
1816 if isinstance(res, Leaf):
1820 return list(res.leaves())[-1]
1829 def child_towards(ancestor: Node, descendant: LN) -> Optional[LN]:
1830 """Return the child of `ancestor` that contains `descendant`."""
1831 node: Optional[LN] = descendant
1832 while node and node.parent != ancestor:
1837 def container_of(leaf: Leaf) -> LN:
1838 """Return `leaf` or one of its ancestors that is the topmost container of it.
1840 By "container" we mean a node where `leaf` is the very first child.
1842 same_prefix = leaf.prefix
1843 container: LN = leaf
1845 parent = container.parent
1849 if parent.children[0].prefix != same_prefix:
1852 if parent.type == syms.file_input:
1855 if parent.type in SURROUNDED_BY_BRACKETS:
1862 def is_split_after_delimiter(leaf: Leaf, previous: Leaf = None) -> int:
1863 """Return the priority of the `leaf` delimiter, given a line break after it.
1865 The delimiter priorities returned here are from those delimiters that would
1866 cause a line break after themselves.
1868 Higher numbers are higher priority.
1870 if leaf.type == token.COMMA:
1871 return COMMA_PRIORITY
1876 def is_split_before_delimiter(leaf: Leaf, previous: Leaf = None) -> int:
1877 """Return the priority of the `leaf` delimiter, given a line before after it.
1879 The delimiter priorities returned here are from those delimiters that would
1880 cause a line break before themselves.
1882 Higher numbers are higher priority.
1884 if is_vararg(leaf, within=VARARGS_PARENTS | UNPACKING_PARENTS):
1885 # * and ** might also be MATH_OPERATORS but in this case they are not.
1886 # Don't treat them as a delimiter.
1890 leaf.type == token.DOT
1892 and leaf.parent.type not in {syms.import_from, syms.dotted_name}
1893 and (previous is None or previous.type in CLOSING_BRACKETS)
1898 leaf.type in MATH_OPERATORS
1900 and leaf.parent.type not in {syms.factor, syms.star_expr}
1902 return MATH_PRIORITIES[leaf.type]
1904 if leaf.type in COMPARATORS:
1905 return COMPARATOR_PRIORITY
1908 leaf.type == token.STRING
1909 and previous is not None
1910 and previous.type == token.STRING
1912 return STRING_PRIORITY
1914 if leaf.type != token.NAME:
1920 and leaf.parent.type in {syms.comp_for, syms.old_comp_for}
1922 return COMPREHENSION_PRIORITY
1927 and leaf.parent.type in {syms.comp_if, syms.old_comp_if}
1929 return COMPREHENSION_PRIORITY
1931 if leaf.value in {"if", "else"} and leaf.parent and leaf.parent.type == syms.test:
1932 return TERNARY_PRIORITY
1934 if leaf.value == "is":
1935 return COMPARATOR_PRIORITY
1940 and leaf.parent.type in {syms.comp_op, syms.comparison}
1942 previous is not None
1943 and previous.type == token.NAME
1944 and previous.value == "not"
1947 return COMPARATOR_PRIORITY
1952 and leaf.parent.type == syms.comp_op
1954 previous is not None
1955 and previous.type == token.NAME
1956 and previous.value == "is"
1959 return COMPARATOR_PRIORITY
1961 if leaf.value in LOGIC_OPERATORS and leaf.parent:
1962 return LOGIC_PRIORITY
1967 FMT_OFF = {"# fmt: off", "# fmt:off", "# yapf: disable"}
1968 FMT_ON = {"# fmt: on", "# fmt:on", "# yapf: enable"}
1971 def generate_comments(leaf: LN) -> Iterator[Leaf]:
1972 """Clean the prefix of the `leaf` and generate comments from it, if any.
1974 Comments in lib2to3 are shoved into the whitespace prefix. This happens
1975 in `pgen2/driver.py:Driver.parse_tokens()`. This was a brilliant implementation
1976 move because it does away with modifying the grammar to include all the
1977 possible places in which comments can be placed.
1979 The sad consequence for us though is that comments don't "belong" anywhere.
1980 This is why this function generates simple parentless Leaf objects for
1981 comments. We simply don't know what the correct parent should be.
1983 No matter though, we can live without this. We really only need to
1984 differentiate between inline and standalone comments. The latter don't
1985 share the line with any code.
1987 Inline comments are emitted as regular token.COMMENT leaves. Standalone
1988 are emitted with a fake STANDALONE_COMMENT token identifier.
1990 for pc in list_comments(leaf.prefix, is_endmarker=leaf.type == token.ENDMARKER):
1991 yield Leaf(pc.type, pc.value, prefix="\n" * pc.newlines)
1996 type: int # token.COMMENT or STANDALONE_COMMENT
1997 value: str # content of the comment
1998 newlines: int # how many newlines before the comment
1999 consumed: int # how many characters of the original leaf's prefix did we consume
2002 @lru_cache(maxsize=4096)
2003 def list_comments(prefix: str, *, is_endmarker: bool) -> List[ProtoComment]:
2004 result: List[ProtoComment] = []
2005 if not prefix or "#" not in prefix:
2010 for index, line in enumerate(prefix.split("\n")):
2011 consumed += len(line) + 1 # adding the length of the split '\n'
2012 line = line.lstrip()
2015 if not line.startswith("#"):
2018 if index == 0 and not is_endmarker:
2019 comment_type = token.COMMENT # simple trailing comment
2021 comment_type = STANDALONE_COMMENT
2022 comment = make_comment(line)
2025 type=comment_type, value=comment, newlines=nlines, consumed=consumed
2032 def make_comment(content: str) -> str:
2033 """Return a consistently formatted comment from the given `content` string.
2035 All comments (except for "##", "#!", "#:") should have a single space between
2036 the hash sign and the content.
2038 If `content` didn't start with a hash sign, one is provided.
2040 content = content.rstrip()
2044 if content[0] == "#":
2045 content = content[1:]
2046 if content and content[0] not in " !:#":
2047 content = " " + content
2048 return "#" + content
2052 line: Line, line_length: int, inner: bool = False, py36: bool = False
2053 ) -> Iterator[Line]:
2054 """Split a `line` into potentially many lines.
2056 They should fit in the allotted `line_length` but might not be able to.
2057 `inner` signifies that there were a pair of brackets somewhere around the
2058 current `line`, possibly transitively. This means we can fallback to splitting
2059 by delimiters if the LHS/RHS don't yield any results.
2061 If `py36` is True, splitting may generate syntax that is only compatible
2062 with Python 3.6 and later.
2068 line_str = str(line).strip("\n")
2069 if not line.should_explode and is_line_short_enough(
2070 line, line_length=line_length, line_str=line_str
2075 split_funcs: List[SplitFunc]
2077 split_funcs = [left_hand_split]
2080 def rhs(line: Line, py36: bool = False) -> Iterator[Line]:
2081 for omit in generate_trailers_to_omit(line, line_length):
2082 lines = list(right_hand_split(line, line_length, py36, omit=omit))
2083 if is_line_short_enough(lines[0], line_length=line_length):
2087 # All splits failed, best effort split with no omits.
2088 # This mostly happens to multiline strings that are by definition
2089 # reported as not fitting a single line.
2090 yield from right_hand_split(line, py36)
2092 if line.inside_brackets:
2093 split_funcs = [delimiter_split, standalone_comment_split, rhs]
2096 for split_func in split_funcs:
2097 # We are accumulating lines in `result` because we might want to abort
2098 # mission and return the original line in the end, or attempt a different
2100 result: List[Line] = []
2102 for l in split_func(line, py36):
2103 if str(l).strip("\n") == line_str:
2104 raise CannotSplit("Split function returned an unchanged result")
2107 split_line(l, line_length=line_length, inner=True, py36=py36)
2109 except CannotSplit as cs:
2120 def left_hand_split(line: Line, py36: bool = False) -> Iterator[Line]:
2121 """Split line into many lines, starting with the first matching bracket pair.
2123 Note: this usually looks weird, only use this for function definitions.
2124 Prefer RHS otherwise. This is why this function is not symmetrical with
2125 :func:`right_hand_split` which also handles optional parentheses.
2127 head = Line(depth=line.depth)
2128 body = Line(depth=line.depth + 1, inside_brackets=True)
2129 tail = Line(depth=line.depth)
2130 tail_leaves: List[Leaf] = []
2131 body_leaves: List[Leaf] = []
2132 head_leaves: List[Leaf] = []
2133 current_leaves = head_leaves
2134 matching_bracket = None
2135 for leaf in line.leaves:
2137 current_leaves is body_leaves
2138 and leaf.type in CLOSING_BRACKETS
2139 and leaf.opening_bracket is matching_bracket
2141 current_leaves = tail_leaves if body_leaves else head_leaves
2142 current_leaves.append(leaf)
2143 if current_leaves is head_leaves:
2144 if leaf.type in OPENING_BRACKETS:
2145 matching_bracket = leaf
2146 current_leaves = body_leaves
2147 # Since body is a new indent level, remove spurious leading whitespace.
2149 normalize_prefix(body_leaves[0], inside_brackets=True)
2150 # Build the new lines.
2151 for result, leaves in (head, head_leaves), (body, body_leaves), (tail, tail_leaves):
2153 result.append(leaf, preformatted=True)
2154 for comment_after in line.comments_after(leaf):
2155 result.append(comment_after, preformatted=True)
2156 bracket_split_succeeded_or_raise(head, body, tail)
2157 for result in (head, body, tail):
2162 def right_hand_split(
2163 line: Line, line_length: int, py36: bool = False, omit: Collection[LeafID] = ()
2164 ) -> Iterator[Line]:
2165 """Split line into many lines, starting with the last matching bracket pair.
2167 If the split was by optional parentheses, attempt splitting without them, too.
2168 `omit` is a collection of closing bracket IDs that shouldn't be considered for
2171 Note: running this function modifies `bracket_depth` on the leaves of `line`.
2173 head = Line(depth=line.depth)
2174 body = Line(depth=line.depth + 1, inside_brackets=True)
2175 tail = Line(depth=line.depth)
2176 tail_leaves: List[Leaf] = []
2177 body_leaves: List[Leaf] = []
2178 head_leaves: List[Leaf] = []
2179 current_leaves = tail_leaves
2180 opening_bracket = None
2181 closing_bracket = None
2182 for leaf in reversed(line.leaves):
2183 if current_leaves is body_leaves:
2184 if leaf is opening_bracket:
2185 current_leaves = head_leaves if body_leaves else tail_leaves
2186 current_leaves.append(leaf)
2187 if current_leaves is tail_leaves:
2188 if leaf.type in CLOSING_BRACKETS and id(leaf) not in omit:
2189 opening_bracket = leaf.opening_bracket
2190 closing_bracket = leaf
2191 current_leaves = body_leaves
2192 tail_leaves.reverse()
2193 body_leaves.reverse()
2194 head_leaves.reverse()
2195 # Since body is a new indent level, remove spurious leading whitespace.
2197 normalize_prefix(body_leaves[0], inside_brackets=True)
2199 # No `head` means the split failed. Either `tail` has all content or
2200 # the matching `opening_bracket` wasn't available on `line` anymore.
2201 raise CannotSplit("No brackets found")
2203 # Build the new lines.
2204 for result, leaves in (head, head_leaves), (body, body_leaves), (tail, tail_leaves):
2206 result.append(leaf, preformatted=True)
2207 for comment_after in line.comments_after(leaf):
2208 result.append(comment_after, preformatted=True)
2209 assert opening_bracket and closing_bracket
2210 body.should_explode = should_explode(body, opening_bracket)
2211 bracket_split_succeeded_or_raise(head, body, tail)
2213 # the body shouldn't be exploded
2214 not body.should_explode
2215 # the opening bracket is an optional paren
2216 and opening_bracket.type == token.LPAR
2217 and not opening_bracket.value
2218 # the closing bracket is an optional paren
2219 and closing_bracket.type == token.RPAR
2220 and not closing_bracket.value
2221 # it's not an import (optional parens are the only thing we can split on
2222 # in this case; attempting a split without them is a waste of time)
2223 and not line.is_import
2224 # there are no standalone comments in the body
2225 and not body.contains_standalone_comments(0)
2226 # and we can actually remove the parens
2227 and can_omit_invisible_parens(body, line_length)
2229 omit = {id(closing_bracket), *omit}
2231 yield from right_hand_split(line, line_length, py36=py36, omit=omit)
2237 or is_line_short_enough(body, line_length=line_length)
2240 "Splitting failed, body is still too long and can't be split."
2243 elif head.contains_multiline_strings() or tail.contains_multiline_strings():
2245 "The current optional pair of parentheses is bound to fail to "
2246 "satisfy the splitting algorithm because the head or the tail "
2247 "contains multiline strings which by definition never fit one "
2251 ensure_visible(opening_bracket)
2252 ensure_visible(closing_bracket)
2253 for result in (head, body, tail):
2258 def bracket_split_succeeded_or_raise(head: Line, body: Line, tail: Line) -> None:
2259 """Raise :exc:`CannotSplit` if the last left- or right-hand split failed.
2261 Do nothing otherwise.
2263 A left- or right-hand split is based on a pair of brackets. Content before
2264 (and including) the opening bracket is left on one line, content inside the
2265 brackets is put on a separate line, and finally content starting with and
2266 following the closing bracket is put on a separate line.
2268 Those are called `head`, `body`, and `tail`, respectively. If the split
2269 produced the same line (all content in `head`) or ended up with an empty `body`
2270 and the `tail` is just the closing bracket, then it's considered failed.
2272 tail_len = len(str(tail).strip())
2275 raise CannotSplit("Splitting brackets produced the same line")
2279 f"Splitting brackets on an empty body to save "
2280 f"{tail_len} characters is not worth it"
2284 def dont_increase_indentation(split_func: SplitFunc) -> SplitFunc:
2285 """Normalize prefix of the first leaf in every line returned by `split_func`.
2287 This is a decorator over relevant split functions.
2291 def split_wrapper(line: Line, py36: bool = False) -> Iterator[Line]:
2292 for l in split_func(line, py36):
2293 normalize_prefix(l.leaves[0], inside_brackets=True)
2296 return split_wrapper
2299 @dont_increase_indentation
2300 def delimiter_split(line: Line, py36: bool = False) -> Iterator[Line]:
2301 """Split according to delimiters of the highest priority.
2303 If `py36` is True, the split will add trailing commas also in function
2304 signatures that contain `*` and `**`.
2307 last_leaf = line.leaves[-1]
2309 raise CannotSplit("Line empty")
2311 bt = line.bracket_tracker
2313 delimiter_priority = bt.max_delimiter_priority(exclude={id(last_leaf)})
2315 raise CannotSplit("No delimiters found")
2317 if delimiter_priority == DOT_PRIORITY:
2318 if bt.delimiter_count_with_priority(delimiter_priority) == 1:
2319 raise CannotSplit("Splitting a single attribute from its owner looks wrong")
2321 current_line = Line(depth=line.depth, inside_brackets=line.inside_brackets)
2322 lowest_depth = sys.maxsize
2323 trailing_comma_safe = True
2325 def append_to_line(leaf: Leaf) -> Iterator[Line]:
2326 """Append `leaf` to current line or to new line if appending impossible."""
2327 nonlocal current_line
2329 current_line.append_safe(leaf, preformatted=True)
2330 except ValueError as ve:
2333 current_line = Line(depth=line.depth, inside_brackets=line.inside_brackets)
2334 current_line.append(leaf)
2336 for index, leaf in enumerate(line.leaves):
2337 yield from append_to_line(leaf)
2339 for comment_after in line.comments_after(leaf, index):
2340 yield from append_to_line(comment_after)
2342 lowest_depth = min(lowest_depth, leaf.bracket_depth)
2343 if leaf.bracket_depth == lowest_depth and is_vararg(
2344 leaf, within=VARARGS_PARENTS
2346 trailing_comma_safe = trailing_comma_safe and py36
2347 leaf_priority = bt.delimiters.get(id(leaf))
2348 if leaf_priority == delimiter_priority:
2351 current_line = Line(depth=line.depth, inside_brackets=line.inside_brackets)
2355 and delimiter_priority == COMMA_PRIORITY
2356 and current_line.leaves[-1].type != token.COMMA
2357 and current_line.leaves[-1].type != STANDALONE_COMMENT
2359 current_line.append(Leaf(token.COMMA, ","))
2363 @dont_increase_indentation
2364 def standalone_comment_split(line: Line, py36: bool = False) -> Iterator[Line]:
2365 """Split standalone comments from the rest of the line."""
2366 if not line.contains_standalone_comments(0):
2367 raise CannotSplit("Line does not have any standalone comments")
2369 current_line = Line(depth=line.depth, inside_brackets=line.inside_brackets)
2371 def append_to_line(leaf: Leaf) -> Iterator[Line]:
2372 """Append `leaf` to current line or to new line if appending impossible."""
2373 nonlocal current_line
2375 current_line.append_safe(leaf, preformatted=True)
2376 except ValueError as ve:
2379 current_line = Line(depth=line.depth, inside_brackets=line.inside_brackets)
2380 current_line.append(leaf)
2382 for index, leaf in enumerate(line.leaves):
2383 yield from append_to_line(leaf)
2385 for comment_after in line.comments_after(leaf, index):
2386 yield from append_to_line(comment_after)
2392 def is_import(leaf: Leaf) -> bool:
2393 """Return True if the given leaf starts an import statement."""
2400 (v == "import" and p and p.type == syms.import_name)
2401 or (v == "from" and p and p.type == syms.import_from)
2406 def normalize_prefix(leaf: Leaf, *, inside_brackets: bool) -> None:
2407 """Leave existing extra newlines if not `inside_brackets`. Remove everything
2410 Note: don't use backslashes for formatting or you'll lose your voting rights.
2412 if not inside_brackets:
2413 spl = leaf.prefix.split("#")
2414 if "\\" not in spl[0]:
2415 nl_count = spl[-1].count("\n")
2418 leaf.prefix = "\n" * nl_count
2424 def normalize_string_prefix(leaf: Leaf, remove_u_prefix: bool = False) -> None:
2425 """Make all string prefixes lowercase.
2427 If remove_u_prefix is given, also removes any u prefix from the string.
2429 Note: Mutates its argument.
2431 match = re.match(r"^([furbFURB]*)(.*)$", leaf.value, re.DOTALL)
2432 assert match is not None, f"failed to match string {leaf.value!r}"
2433 orig_prefix = match.group(1)
2434 new_prefix = orig_prefix.lower()
2436 new_prefix = new_prefix.replace("u", "")
2437 leaf.value = f"{new_prefix}{match.group(2)}"
2440 def normalize_string_quotes(leaf: Leaf) -> None:
2441 """Prefer double quotes but only if it doesn't cause more escaping.
2443 Adds or removes backslashes as appropriate. Doesn't parse and fix
2444 strings nested in f-strings (yet).
2446 Note: Mutates its argument.
2448 value = leaf.value.lstrip("furbFURB")
2449 if value[:3] == '"""':
2452 elif value[:3] == "'''":
2455 elif value[0] == '"':
2461 first_quote_pos = leaf.value.find(orig_quote)
2462 if first_quote_pos == -1:
2463 return # There's an internal error
2465 prefix = leaf.value[:first_quote_pos]
2466 unescaped_new_quote = re.compile(rf"(([^\\]|^)(\\\\)*){new_quote}")
2467 escaped_new_quote = re.compile(rf"([^\\]|^)\\((?:\\\\)*){new_quote}")
2468 escaped_orig_quote = re.compile(rf"([^\\]|^)\\((?:\\\\)*){orig_quote}")
2469 body = leaf.value[first_quote_pos + len(orig_quote) : -len(orig_quote)]
2470 if "r" in prefix.casefold():
2471 if unescaped_new_quote.search(body):
2472 # There's at least one unescaped new_quote in this raw string
2473 # so converting is impossible
2476 # Do not introduce or remove backslashes in raw strings
2479 # remove unnecessary escapes
2480 new_body = sub_twice(escaped_new_quote, rf"\1\2{new_quote}", body)
2481 if body != new_body:
2482 # Consider the string without unnecessary escapes as the original
2484 leaf.value = f"{prefix}{orig_quote}{body}{orig_quote}"
2485 new_body = sub_twice(escaped_orig_quote, rf"\1\2{orig_quote}", new_body)
2486 new_body = sub_twice(unescaped_new_quote, rf"\1\\{new_quote}", new_body)
2487 if "f" in prefix.casefold():
2488 matches = re.findall(r"[^{]\{(.*?)\}[^}]", new_body)
2491 # Do not introduce backslashes in interpolated expressions
2493 if new_quote == '"""' and new_body[-1:] == '"':
2495 new_body = new_body[:-1] + '\\"'
2496 orig_escape_count = body.count("\\")
2497 new_escape_count = new_body.count("\\")
2498 if new_escape_count > orig_escape_count:
2499 return # Do not introduce more escaping
2501 if new_escape_count == orig_escape_count and orig_quote == '"':
2502 return # Prefer double quotes
2504 leaf.value = f"{prefix}{new_quote}{new_body}{new_quote}"
2507 def normalize_invisible_parens(node: Node, parens_after: Set[str]) -> None:
2508 """Make existing optional parentheses invisible or create new ones.
2510 `parens_after` is a set of string leaf values immeditely after which parens
2513 Standardizes on visible parentheses for single-element tuples, and keeps
2514 existing visible parentheses for other tuples and generator expressions.
2516 for pc in list_comments(node.prefix, is_endmarker=False):
2517 if pc.value in FMT_OFF:
2518 # This `node` has a prefix with `# fmt: off`, don't mess with parens.
2522 for index, child in enumerate(list(node.children)):
2524 if child.type == syms.atom:
2525 maybe_make_parens_invisible_in_atom(child)
2526 elif is_one_tuple(child):
2527 # wrap child in visible parentheses
2528 lpar = Leaf(token.LPAR, "(")
2529 rpar = Leaf(token.RPAR, ")")
2531 node.insert_child(index, Node(syms.atom, [lpar, child, rpar]))
2532 elif node.type == syms.import_from:
2533 # "import from" nodes store parentheses directly as part of
2535 if child.type == token.LPAR:
2536 # make parentheses invisible
2537 child.value = "" # type: ignore
2538 node.children[-1].value = "" # type: ignore
2539 elif child.type != token.STAR:
2540 # insert invisible parentheses
2541 node.insert_child(index, Leaf(token.LPAR, ""))
2542 node.append_child(Leaf(token.RPAR, ""))
2545 elif not (isinstance(child, Leaf) and is_multiline_string(child)):
2546 # wrap child in invisible parentheses
2547 lpar = Leaf(token.LPAR, "")
2548 rpar = Leaf(token.RPAR, "")
2549 index = child.remove() or 0
2550 node.insert_child(index, Node(syms.atom, [lpar, child, rpar]))
2552 check_lpar = isinstance(child, Leaf) and child.value in parens_after
2555 def normalize_fmt_off(node: Node) -> None:
2556 """Convert content between `# fmt: off`/`# fmt: on` into standalone comments."""
2559 try_again = convert_one_fmt_off_pair(node)
2562 def convert_one_fmt_off_pair(node: Node) -> bool:
2563 """Convert content of a single `# fmt: off`/`# fmt: on` into a standalone comment.
2565 Returns True if a pair was converted.
2567 for leaf in node.leaves():
2568 previous_consumed = 0
2569 for comment in list_comments(leaf.prefix, is_endmarker=False):
2570 if comment.value in FMT_OFF:
2571 # We only want standalone comments. If there's no previous leaf or
2572 # the previous leaf is indentation, it's a standalone comment in
2574 if comment.type != STANDALONE_COMMENT:
2575 prev = preceding_leaf(leaf)
2576 if prev and prev.type not in WHITESPACE:
2579 ignored_nodes = list(generate_ignored_nodes(leaf))
2580 if not ignored_nodes:
2583 first = ignored_nodes[0] # Can be a container node with the `leaf`.
2584 parent = first.parent
2585 prefix = first.prefix
2586 first.prefix = prefix[comment.consumed :]
2588 comment.value + "\n" + "".join(str(n) for n in ignored_nodes)
2590 if hidden_value.endswith("\n"):
2591 # That happens when one of the `ignored_nodes` ended with a NEWLINE
2592 # leaf (possibly followed by a DEDENT).
2593 hidden_value = hidden_value[:-1]
2595 for ignored in ignored_nodes:
2596 index = ignored.remove()
2597 if first_idx is None:
2599 assert parent is not None, "INTERNAL ERROR: fmt: on/off handling (1)"
2600 assert first_idx is not None, "INTERNAL ERROR: fmt: on/off handling (2)"
2601 parent.insert_child(
2606 prefix=prefix[:previous_consumed] + "\n" * comment.newlines,
2611 previous_consumed += comment.consumed
2616 def generate_ignored_nodes(leaf: Leaf) -> Iterator[LN]:
2617 """Starting from the container of `leaf`, generate all leaves until `# fmt: on`.
2619 Stops at the end of the block.
2621 container: Optional[LN] = container_of(leaf)
2622 while container is not None and container.type != token.ENDMARKER:
2623 for comment in list_comments(container.prefix, is_endmarker=False):
2624 if comment.value in FMT_ON:
2629 container = container.next_sibling
2632 def maybe_make_parens_invisible_in_atom(node: LN) -> bool:
2633 """If it's safe, make the parens in the atom `node` invisible, recursively."""
2635 node.type != syms.atom
2636 or is_empty_tuple(node)
2637 or is_one_tuple(node)
2639 or max_delimiter_priority_in_atom(node) >= COMMA_PRIORITY
2643 first = node.children[0]
2644 last = node.children[-1]
2645 if first.type == token.LPAR and last.type == token.RPAR:
2646 # make parentheses invisible
2647 first.value = "" # type: ignore
2648 last.value = "" # type: ignore
2649 if len(node.children) > 1:
2650 maybe_make_parens_invisible_in_atom(node.children[1])
2656 def is_empty_tuple(node: LN) -> bool:
2657 """Return True if `node` holds an empty tuple."""
2659 node.type == syms.atom
2660 and len(node.children) == 2
2661 and node.children[0].type == token.LPAR
2662 and node.children[1].type == token.RPAR
2666 def is_one_tuple(node: LN) -> bool:
2667 """Return True if `node` holds a tuple with one element, with or without parens."""
2668 if node.type == syms.atom:
2669 if len(node.children) != 3:
2672 lpar, gexp, rpar = node.children
2674 lpar.type == token.LPAR
2675 and gexp.type == syms.testlist_gexp
2676 and rpar.type == token.RPAR
2680 return len(gexp.children) == 2 and gexp.children[1].type == token.COMMA
2683 node.type in IMPLICIT_TUPLE
2684 and len(node.children) == 2
2685 and node.children[1].type == token.COMMA
2689 def is_yield(node: LN) -> bool:
2690 """Return True if `node` holds a `yield` or `yield from` expression."""
2691 if node.type == syms.yield_expr:
2694 if node.type == token.NAME and node.value == "yield": # type: ignore
2697 if node.type != syms.atom:
2700 if len(node.children) != 3:
2703 lpar, expr, rpar = node.children
2704 if lpar.type == token.LPAR and rpar.type == token.RPAR:
2705 return is_yield(expr)
2710 def is_vararg(leaf: Leaf, within: Set[NodeType]) -> bool:
2711 """Return True if `leaf` is a star or double star in a vararg or kwarg.
2713 If `within` includes VARARGS_PARENTS, this applies to function signatures.
2714 If `within` includes UNPACKING_PARENTS, it applies to right hand-side
2715 extended iterable unpacking (PEP 3132) and additional unpacking
2716 generalizations (PEP 448).
2718 if leaf.type not in STARS or not leaf.parent:
2722 if p.type == syms.star_expr:
2723 # Star expressions are also used as assignment targets in extended
2724 # iterable unpacking (PEP 3132). See what its parent is instead.
2730 return p.type in within
2733 def is_multiline_string(leaf: Leaf) -> bool:
2734 """Return True if `leaf` is a multiline string that actually spans many lines."""
2735 value = leaf.value.lstrip("furbFURB")
2736 return value[:3] in {'"""', "'''"} and "\n" in value
2739 def is_stub_suite(node: Node) -> bool:
2740 """Return True if `node` is a suite with a stub body."""
2742 len(node.children) != 4
2743 or node.children[0].type != token.NEWLINE
2744 or node.children[1].type != token.INDENT
2745 or node.children[3].type != token.DEDENT
2749 return is_stub_body(node.children[2])
2752 def is_stub_body(node: LN) -> bool:
2753 """Return True if `node` is a simple statement containing an ellipsis."""
2754 if not isinstance(node, Node) or node.type != syms.simple_stmt:
2757 if len(node.children) != 2:
2760 child = node.children[0]
2762 child.type == syms.atom
2763 and len(child.children) == 3
2764 and all(leaf == Leaf(token.DOT, ".") for leaf in child.children)
2768 def max_delimiter_priority_in_atom(node: LN) -> int:
2769 """Return maximum delimiter priority inside `node`.
2771 This is specific to atoms with contents contained in a pair of parentheses.
2772 If `node` isn't an atom or there are no enclosing parentheses, returns 0.
2774 if node.type != syms.atom:
2777 first = node.children[0]
2778 last = node.children[-1]
2779 if not (first.type == token.LPAR and last.type == token.RPAR):
2782 bt = BracketTracker()
2783 for c in node.children[1:-1]:
2784 if isinstance(c, Leaf):
2787 for leaf in c.leaves():
2790 return bt.max_delimiter_priority()
2796 def ensure_visible(leaf: Leaf) -> None:
2797 """Make sure parentheses are visible.
2799 They could be invisible as part of some statements (see
2800 :func:`normalize_invible_parens` and :func:`visit_import_from`).
2802 if leaf.type == token.LPAR:
2804 elif leaf.type == token.RPAR:
2808 def should_explode(line: Line, opening_bracket: Leaf) -> bool:
2809 """Should `line` immediately be split with `delimiter_split()` after RHS?"""
2811 opening_bracket.parent
2812 and opening_bracket.parent.type in {syms.atom, syms.import_from}
2813 and opening_bracket.value in "[{("
2818 last_leaf = line.leaves[-1]
2819 exclude = {id(last_leaf)} if last_leaf.type == token.COMMA else set()
2820 max_priority = line.bracket_tracker.max_delimiter_priority(exclude=exclude)
2821 except (IndexError, ValueError):
2824 return max_priority == COMMA_PRIORITY
2827 def is_python36(node: Node) -> bool:
2828 """Return True if the current file is using Python 3.6+ features.
2830 Currently looking for:
2832 - trailing commas after * or ** in function signatures and calls.
2834 for n in node.pre_order():
2835 if n.type == token.STRING:
2836 value_head = n.value[:2] # type: ignore
2837 if value_head in {'f"', 'F"', "f'", "F'", "rf", "fr", "RF", "FR"}:
2841 n.type in {syms.typedargslist, syms.arglist}
2843 and n.children[-1].type == token.COMMA
2845 for ch in n.children:
2846 if ch.type in STARS:
2849 if ch.type == syms.argument:
2850 for argch in ch.children:
2851 if argch.type in STARS:
2857 def generate_trailers_to_omit(line: Line, line_length: int) -> Iterator[Set[LeafID]]:
2858 """Generate sets of closing bracket IDs that should be omitted in a RHS.
2860 Brackets can be omitted if the entire trailer up to and including
2861 a preceding closing bracket fits in one line.
2863 Yielded sets are cumulative (contain results of previous yields, too). First
2867 omit: Set[LeafID] = set()
2870 length = 4 * line.depth
2871 opening_bracket = None
2872 closing_bracket = None
2873 optional_brackets: Set[LeafID] = set()
2874 inner_brackets: Set[LeafID] = set()
2875 for index, leaf, leaf_length in enumerate_with_length(line, reversed=True):
2876 length += leaf_length
2877 if length > line_length:
2880 has_inline_comment = leaf_length > len(leaf.value) + len(leaf.prefix)
2881 if leaf.type == STANDALONE_COMMENT or has_inline_comment:
2884 optional_brackets.discard(id(leaf))
2886 if leaf is opening_bracket:
2887 opening_bracket = None
2888 elif leaf.type in CLOSING_BRACKETS:
2889 inner_brackets.add(id(leaf))
2890 elif leaf.type in CLOSING_BRACKETS:
2892 optional_brackets.add(id(opening_bracket))
2895 if index > 0 and line.leaves[index - 1].type in OPENING_BRACKETS:
2896 # Empty brackets would fail a split so treat them as "inner"
2897 # brackets (e.g. only add them to the `omit` set if another
2898 # pair of brackets was good enough.
2899 inner_brackets.add(id(leaf))
2902 opening_bracket = leaf.opening_bracket
2904 omit.add(id(closing_bracket))
2905 omit.update(inner_brackets)
2906 inner_brackets.clear()
2908 closing_bracket = leaf
2911 def get_future_imports(node: Node) -> Set[str]:
2912 """Return a set of __future__ imports in the file."""
2914 for child in node.children:
2915 if child.type != syms.simple_stmt:
2917 first_child = child.children[0]
2918 if isinstance(first_child, Leaf):
2919 # Continue looking if we see a docstring; otherwise stop.
2921 len(child.children) == 2
2922 and first_child.type == token.STRING
2923 and child.children[1].type == token.NEWLINE
2928 elif first_child.type == syms.import_from:
2929 module_name = first_child.children[1]
2930 if not isinstance(module_name, Leaf) or module_name.value != "__future__":
2932 for import_from_child in first_child.children[3:]:
2933 if isinstance(import_from_child, Leaf):
2934 if import_from_child.type == token.NAME:
2935 imports.add(import_from_child.value)
2937 assert import_from_child.type == syms.import_as_names
2938 for leaf in import_from_child.children:
2939 if isinstance(leaf, Leaf) and leaf.type == token.NAME:
2940 imports.add(leaf.value)
2946 def gen_python_files_in_dir(
2949 include: Pattern[str],
2950 exclude: Pattern[str],
2952 ) -> Iterator[Path]:
2953 """Generate all files under `path` whose paths are not excluded by the
2954 `exclude` regex, but are included by the `include` regex.
2956 Symbolic links pointing outside of the `root` directory are ignored.
2958 `report` is where output about exclusions goes.
2960 assert root.is_absolute(), f"INTERNAL ERROR: `root` must be absolute but is {root}"
2961 for child in path.iterdir():
2963 normalized_path = "/" + child.resolve().relative_to(root).as_posix()
2965 if child.is_symlink():
2966 report.path_ignored(
2967 child, f"is a symbolic link that points outside {root}"
2974 normalized_path += "/"
2975 exclude_match = exclude.search(normalized_path)
2976 if exclude_match and exclude_match.group(0):
2977 report.path_ignored(child, f"matches the --exclude regular expression")
2981 yield from gen_python_files_in_dir(child, root, include, exclude, report)
2983 elif child.is_file():
2984 include_match = include.search(normalized_path)
2990 def find_project_root(srcs: Iterable[str]) -> Path:
2991 """Return a directory containing .git, .hg, or pyproject.toml.
2993 That directory can be one of the directories passed in `srcs` or their
2996 If no directory in the tree contains a marker that would specify it's the
2997 project root, the root of the file system is returned.
3000 return Path("/").resolve()
3002 common_base = min(Path(src).resolve() for src in srcs)
3003 if common_base.is_dir():
3004 # Append a fake file so `parents` below returns `common_base_dir`, too.
3005 common_base /= "fake-file"
3006 for directory in common_base.parents:
3007 if (directory / ".git").is_dir():
3010 if (directory / ".hg").is_dir():
3013 if (directory / "pyproject.toml").is_file():
3021 """Provides a reformatting counter. Can be rendered with `str(report)`."""
3025 verbose: bool = False
3026 change_count: int = 0
3028 failure_count: int = 0
3030 def done(self, src: Path, changed: Changed) -> None:
3031 """Increment the counter for successful reformatting. Write out a message."""
3032 if changed is Changed.YES:
3033 reformatted = "would reformat" if self.check else "reformatted"
3034 if self.verbose or not self.quiet:
3035 out(f"{reformatted} {src}")
3036 self.change_count += 1
3039 if changed is Changed.NO:
3040 msg = f"{src} already well formatted, good job."
3042 msg = f"{src} wasn't modified on disk since last run."
3043 out(msg, bold=False)
3044 self.same_count += 1
3046 def failed(self, src: Path, message: str) -> None:
3047 """Increment the counter for failed reformatting. Write out a message."""
3048 err(f"error: cannot format {src}: {message}")
3049 self.failure_count += 1
3051 def path_ignored(self, path: Path, message: str) -> None:
3053 out(f"{path} ignored: {message}", bold=False)
3056 def return_code(self) -> int:
3057 """Return the exit code that the app should use.
3059 This considers the current state of changed files and failures:
3060 - if there were any failures, return 123;
3061 - if any files were changed and --check is being used, return 1;
3062 - otherwise return 0.
3064 # According to http://tldp.org/LDP/abs/html/exitcodes.html starting with
3065 # 126 we have special returncodes reserved by the shell.
3066 if self.failure_count:
3069 elif self.change_count and self.check:
3074 def __str__(self) -> str:
3075 """Render a color report of the current state.
3077 Use `click.unstyle` to remove colors.
3080 reformatted = "would be reformatted"
3081 unchanged = "would be left unchanged"
3082 failed = "would fail to reformat"
3084 reformatted = "reformatted"
3085 unchanged = "left unchanged"
3086 failed = "failed to reformat"
3088 if self.change_count:
3089 s = "s" if self.change_count > 1 else ""
3091 click.style(f"{self.change_count} file{s} {reformatted}", bold=True)
3094 s = "s" if self.same_count > 1 else ""
3095 report.append(f"{self.same_count} file{s} {unchanged}")
3096 if self.failure_count:
3097 s = "s" if self.failure_count > 1 else ""
3099 click.style(f"{self.failure_count} file{s} {failed}", fg="red")
3101 return ", ".join(report) + "."
3104 def assert_equivalent(src: str, dst: str) -> None:
3105 """Raise AssertionError if `src` and `dst` aren't equivalent."""
3110 def _v(node: ast.AST, depth: int = 0) -> Iterator[str]:
3111 """Simple visitor generating strings to compare ASTs by content."""
3112 yield f"{' ' * depth}{node.__class__.__name__}("
3114 for field in sorted(node._fields):
3116 value = getattr(node, field)
3117 except AttributeError:
3120 yield f"{' ' * (depth+1)}{field}="
3122 if isinstance(value, list):
3124 if isinstance(item, ast.AST):
3125 yield from _v(item, depth + 2)
3127 elif isinstance(value, ast.AST):
3128 yield from _v(value, depth + 2)
3131 yield f"{' ' * (depth+2)}{value!r}, # {value.__class__.__name__}"
3133 yield f"{' ' * depth}) # /{node.__class__.__name__}"
3136 src_ast = ast.parse(src)
3137 except Exception as exc:
3138 major, minor = sys.version_info[:2]
3139 raise AssertionError(
3140 f"cannot use --safe with this file; failed to parse source file "
3141 f"with Python {major}.{minor}'s builtin AST. Re-run with --fast "
3142 f"or stop using deprecated Python 2 syntax. AST error message: {exc}"
3146 dst_ast = ast.parse(dst)
3147 except Exception as exc:
3148 log = dump_to_file("".join(traceback.format_tb(exc.__traceback__)), dst)
3149 raise AssertionError(
3150 f"INTERNAL ERROR: Black produced invalid code: {exc}. "
3151 f"Please report a bug on https://github.com/ambv/black/issues. "
3152 f"This invalid output might be helpful: {log}"
3155 src_ast_str = "\n".join(_v(src_ast))
3156 dst_ast_str = "\n".join(_v(dst_ast))
3157 if src_ast_str != dst_ast_str:
3158 log = dump_to_file(diff(src_ast_str, dst_ast_str, "src", "dst"))
3159 raise AssertionError(
3160 f"INTERNAL ERROR: Black produced code that is not equivalent to "
3162 f"Please report a bug on https://github.com/ambv/black/issues. "
3163 f"This diff might be helpful: {log}"
3168 src: str, dst: str, line_length: int, mode: FileMode = FileMode.AUTO_DETECT
3170 """Raise AssertionError if `dst` reformats differently the second time."""
3171 newdst = format_str(dst, line_length=line_length, mode=mode)
3174 diff(src, dst, "source", "first pass"),
3175 diff(dst, newdst, "first pass", "second pass"),
3177 raise AssertionError(
3178 f"INTERNAL ERROR: Black produced different code on the second pass "
3179 f"of the formatter. "
3180 f"Please report a bug on https://github.com/ambv/black/issues. "
3181 f"This diff might be helpful: {log}"
3185 def dump_to_file(*output: str) -> str:
3186 """Dump `output` to a temporary file. Return path to the file."""
3189 with tempfile.NamedTemporaryFile(
3190 mode="w", prefix="blk_", suffix=".log", delete=False, encoding="utf8"
3192 for lines in output:
3194 if lines and lines[-1] != "\n":
3199 def diff(a: str, b: str, a_name: str, b_name: str) -> str:
3200 """Return a unified diff string between strings `a` and `b`."""
3203 a_lines = [line + "\n" for line in a.split("\n")]
3204 b_lines = [line + "\n" for line in b.split("\n")]
3206 difflib.unified_diff(a_lines, b_lines, fromfile=a_name, tofile=b_name, n=5)
3210 def cancel(tasks: Iterable[asyncio.Task]) -> None:
3211 """asyncio signal handler that cancels all `tasks` and reports to stderr."""
3217 def shutdown(loop: BaseEventLoop) -> None:
3218 """Cancel all pending tasks on `loop`, wait for them, and close the loop."""
3220 # This part is borrowed from asyncio/runners.py in Python 3.7b2.
3221 to_cancel = [task for task in asyncio.Task.all_tasks(loop) if not task.done()]
3225 for task in to_cancel:
3227 loop.run_until_complete(
3228 asyncio.gather(*to_cancel, loop=loop, return_exceptions=True)
3231 # `concurrent.futures.Future` objects cannot be cancelled once they
3232 # are already running. There might be some when the `shutdown()` happened.
3233 # Silence their logger's spew about the event loop being closed.
3234 cf_logger = logging.getLogger("concurrent.futures")
3235 cf_logger.setLevel(logging.CRITICAL)
3239 def sub_twice(regex: Pattern[str], replacement: str, original: str) -> str:
3240 """Replace `regex` with `replacement` twice on `original`.
3242 This is used by string normalization to perform replaces on
3243 overlapping matches.
3245 return regex.sub(replacement, regex.sub(replacement, original))
3248 def re_compile_maybe_verbose(regex: str) -> Pattern[str]:
3249 """Compile a regular expression string in `regex`.
3251 If it contains newlines, use verbose mode.
3254 regex = "(?x)" + regex
3255 return re.compile(regex)
3258 def enumerate_reversed(sequence: Sequence[T]) -> Iterator[Tuple[Index, T]]:
3259 """Like `reversed(enumerate(sequence))` if that were possible."""
3260 index = len(sequence) - 1
3261 for element in reversed(sequence):
3262 yield (index, element)
3266 def enumerate_with_length(
3267 line: Line, reversed: bool = False
3268 ) -> Iterator[Tuple[Index, Leaf, int]]:
3269 """Return an enumeration of leaves with their length.
3271 Stops prematurely on multiline strings and standalone comments.
3274 Callable[[Sequence[Leaf]], Iterator[Tuple[Index, Leaf]]],
3275 enumerate_reversed if reversed else enumerate,
3277 for index, leaf in op(line.leaves):
3278 length = len(leaf.prefix) + len(leaf.value)
3279 if "\n" in leaf.value:
3280 return # Multiline strings, we can't continue.
3282 comment: Optional[Leaf]
3283 for comment in line.comments_after(leaf, index):
3284 length += len(comment.value)
3286 yield index, leaf, length
3289 def is_line_short_enough(line: Line, *, line_length: int, line_str: str = "") -> bool:
3290 """Return True if `line` is no longer than `line_length`.
3292 Uses the provided `line_str` rendering, if any, otherwise computes a new one.
3295 line_str = str(line).strip("\n")
3297 len(line_str) <= line_length
3298 and "\n" not in line_str # multiline strings
3299 and not line.contains_standalone_comments()
3303 def can_be_split(line: Line) -> bool:
3304 """Return False if the line cannot be split *for sure*.
3306 This is not an exhaustive search but a cheap heuristic that we can use to
3307 avoid some unfortunate formattings (mostly around wrapping unsplittable code
3308 in unnecessary parentheses).
3310 leaves = line.leaves
3314 if leaves[0].type == token.STRING and leaves[1].type == token.DOT:
3318 for leaf in leaves[-2::-1]:
3319 if leaf.type in OPENING_BRACKETS:
3320 if next.type not in CLOSING_BRACKETS:
3324 elif leaf.type == token.DOT:
3326 elif leaf.type == token.NAME:
3327 if not (next.type == token.DOT or next.type in OPENING_BRACKETS):
3330 elif leaf.type not in CLOSING_BRACKETS:
3333 if dot_count > 1 and call_count > 1:
3339 def can_omit_invisible_parens(line: Line, line_length: int) -> bool:
3340 """Does `line` have a shape safe to reformat without optional parens around it?
3342 Returns True for only a subset of potentially nice looking formattings but
3343 the point is to not return false positives that end up producing lines that
3346 bt = line.bracket_tracker
3347 if not bt.delimiters:
3348 # Without delimiters the optional parentheses are useless.
3351 max_priority = bt.max_delimiter_priority()
3352 if bt.delimiter_count_with_priority(max_priority) > 1:
3353 # With more than one delimiter of a kind the optional parentheses read better.
3356 if max_priority == DOT_PRIORITY:
3357 # A single stranded method call doesn't require optional parentheses.
3360 assert len(line.leaves) >= 2, "Stranded delimiter"
3362 first = line.leaves[0]
3363 second = line.leaves[1]
3364 penultimate = line.leaves[-2]
3365 last = line.leaves[-1]
3367 # With a single delimiter, omit if the expression starts or ends with
3369 if first.type in OPENING_BRACKETS and second.type not in CLOSING_BRACKETS:
3371 length = 4 * line.depth
3372 for _index, leaf, leaf_length in enumerate_with_length(line):
3373 if leaf.type in CLOSING_BRACKETS and leaf.opening_bracket is first:
3376 length += leaf_length
3377 if length > line_length:
3380 if leaf.type in OPENING_BRACKETS:
3381 # There are brackets we can further split on.
3385 # checked the entire string and line length wasn't exceeded
3386 if len(line.leaves) == _index + 1:
3389 # Note: we are not returning False here because a line might have *both*
3390 # a leading opening bracket and a trailing closing bracket. If the
3391 # opening bracket doesn't match our rule, maybe the closing will.
3394 last.type == token.RPAR
3395 or last.type == token.RBRACE
3397 # don't use indexing for omitting optional parentheses;
3399 last.type == token.RSQB
3401 and last.parent.type != syms.trailer
3404 if penultimate.type in OPENING_BRACKETS:
3405 # Empty brackets don't help.
3408 if is_multiline_string(first):
3409 # Additional wrapping of a multiline string in this situation is
3413 length = 4 * line.depth
3414 seen_other_brackets = False
3415 for _index, leaf, leaf_length in enumerate_with_length(line):
3416 length += leaf_length
3417 if leaf is last.opening_bracket:
3418 if seen_other_brackets or length <= line_length:
3421 elif leaf.type in OPENING_BRACKETS:
3422 # There are brackets we can further split on.
3423 seen_other_brackets = True
3428 def get_cache_file(line_length: int, mode: FileMode) -> Path:
3429 return CACHE_DIR / f"cache.{line_length}.{mode.value}.pickle"
3432 def read_cache(line_length: int, mode: FileMode) -> Cache:
3433 """Read the cache if it exists and is well formed.
3435 If it is not well formed, the call to write_cache later should resolve the issue.
3437 cache_file = get_cache_file(line_length, mode)
3438 if not cache_file.exists():
3441 with cache_file.open("rb") as fobj:
3443 cache: Cache = pickle.load(fobj)
3444 except pickle.UnpicklingError:
3450 def get_cache_info(path: Path) -> CacheInfo:
3451 """Return the information used to check if a file is already formatted or not."""
3453 return stat.st_mtime, stat.st_size
3456 def filter_cached(cache: Cache, sources: Iterable[Path]) -> Tuple[Set[Path], Set[Path]]:
3457 """Split an iterable of paths in `sources` into two sets.
3459 The first contains paths of files that modified on disk or are not in the
3460 cache. The other contains paths to non-modified files.
3462 todo, done = set(), set()
3465 if cache.get(src) != get_cache_info(src):
3473 cache: Cache, sources: Iterable[Path], line_length: int, mode: FileMode
3475 """Update the cache file."""
3476 cache_file = get_cache_file(line_length, mode)
3478 if not CACHE_DIR.exists():
3479 CACHE_DIR.mkdir(parents=True)
3480 new_cache = {**cache, **{src.resolve(): get_cache_info(src) for src in sources}}
3481 with cache_file.open("wb") as fobj:
3482 pickle.dump(new_cache, fobj, protocol=pickle.HIGHEST_PROTOCOL)
3487 def patch_click() -> None:
3488 """Make Click not crash.
3490 On certain misconfigured environments, Python 3 selects the ASCII encoding as the
3491 default which restricts paths that it can access during the lifetime of the
3492 application. Click refuses to work in this scenario by raising a RuntimeError.
3494 In case of Black the likelihood that non-ASCII characters are going to be used in
3495 file paths is minimal since it's Python source code. Moreover, this crash was
3496 spurious on Python 3.7 thanks to PEP 538 and PEP 540.
3499 from click import core
3500 from click import _unicodefun # type: ignore
3501 except ModuleNotFoundError:
3504 for module in (core, _unicodefun):
3505 if hasattr(module, "_verify_python3_env"):
3506 module._verify_python3_env = lambda: None
3509 if __name__ == "__main__":