X-Git-Url: https://git.madduck.net/etc/vim.git/blobdiff_plain/e1e89091d1de878c98e2d2b0cefd81cc75e7a128..b901d75deb497aa56a1d7651a3386af8a495afc9:/black.py?ds=inline diff --git a/black.py b/black.py index 7ae2627..de86156 100644 --- a/black.py +++ b/black.py @@ -1,4 +1,5 @@ #!/usr/bin/env python3 + import asyncio from asyncio.base_events import BaseEventLoop from concurrent.futures import Executor, ProcessPoolExecutor @@ -7,11 +8,12 @@ import keyword import os from pathlib import Path import tokenize +import sys from typing import ( - Dict, Generic, Iterable, Iterator, List, Optional, Set, Tuple, TypeVar, Union + Dict, Generic, Iterable, Iterator, List, Optional, Set, Tuple, Type, TypeVar, Union ) -from attr import attrib, dataclass, Factory +from attr import dataclass, Factory import click # lib2to3 fork @@ -20,7 +22,7 @@ from blib2to3 import pygram, pytree from blib2to3.pgen2 import driver, token from blib2to3.pgen2.parse import ParseError -__version__ = "18.3a1" +__version__ = "18.3a4" DEFAULT_LINE_LENGTH = 88 # types syms = pygram.python_symbols @@ -36,15 +38,43 @@ err = partial(click.secho, fg='red', err=True) class NothingChanged(UserWarning): - """Raised by `format_file` when the reformatted code is the same as source.""" + """Raised by `format_file()` when the reformatted code is the same as source.""" class CannotSplit(Exception): """A readable split that fits the allotted line length is impossible. - Raised by `left_hand_split()` and `right_hand_split()`. + Raised by `left_hand_split()`, `right_hand_split()`, and `delimiter_split()`. + """ + + +class FormatError(Exception): + """Base exception for `# fmt: on` and `# fmt: off` handling. + + It holds the number of bytes of the prefix consumed before the format + control comment appeared. """ + def __init__(self, consumed: int) -> None: + super().__init__(consumed) + self.consumed = consumed + + def trim_prefix(self, leaf: Leaf) -> None: + leaf.prefix = leaf.prefix[self.consumed:] + + def leaf_from_consumed(self, leaf: Leaf) -> Leaf: + """Returns a new Leaf from the consumed part of the prefix.""" + unformatted_prefix = leaf.prefix[:self.consumed] + return Leaf(token.NEWLINE, unformatted_prefix) + + +class FormatOn(FormatError): + """Found a comment like `# fmt: on` in the file.""" + + +class FormatOff(FormatError): + """Found a comment like `# fmt: off` in the file.""" + @click.command() @click.option( @@ -55,6 +85,15 @@ class CannotSplit(Exception): help='How many character per line to allow.', show_default=True, ) +@click.option( + '--check', + is_flag=True, + help=( + "Don't write back the files, just return the status. Return code 0 " + "means nothing would change. Return code 1 means some files would be " + "reformatted. Return code 123 means there was an internal error." + ), +) @click.option( '--fast/--safe', is_flag=True, @@ -64,10 +103,14 @@ class CannotSplit(Exception): @click.argument( 'src', nargs=-1, - type=click.Path(exists=True, file_okay=True, dir_okay=True, readable=True), + type=click.Path( + exists=True, file_okay=True, dir_okay=True, readable=True, allow_dash=True + ), ) @click.pass_context -def main(ctx: click.Context, line_length: int, fast: bool, src: List[str]) -> None: +def main( + ctx: click.Context, line_length: int, check: bool, fast: bool, src: List[str] +) -> None: """The uncompromising code formatter.""" sources: List[Path] = [] for s in src: @@ -77,15 +120,24 @@ def main(ctx: click.Context, line_length: int, fast: bool, src: List[str]) -> No elif p.is_file(): # if a file was explicitly given, we don't care about its extension sources.append(p) + elif s == '-': + sources.append(Path('-')) else: err(f'invalid path: {s}') if len(sources) == 0: ctx.exit(0) elif len(sources) == 1: p = sources[0] - report = Report() + report = Report(check=check) try: - changed = format_file_in_place(p, line_length=line_length, fast=fast) + if not p.is_file() and str(p) == '-': + changed = format_stdin_to_stdout( + line_length=line_length, fast=fast, write_back=not check + ) + else: + changed = format_file_in_place( + p, line_length=line_length, fast=fast, write_back=not check + ) report.done(p, changed) except Exception as exc: report.failed(p, str(exc)) @@ -96,7 +148,9 @@ def main(ctx: click.Context, line_length: int, fast: bool, src: List[str]) -> No return_code = 1 try: return_code = loop.run_until_complete( - schedule_formatting(sources, line_length, fast, loop, executor) + schedule_formatting( + sources, line_length, not check, fast, loop, executor + ) ) finally: loop.close() @@ -106,13 +160,21 @@ def main(ctx: click.Context, line_length: int, fast: bool, src: List[str]) -> No async def schedule_formatting( sources: List[Path], line_length: int, + write_back: bool, fast: bool, loop: BaseEventLoop, executor: Executor, ) -> int: + """Run formatting of `sources` in parallel using the provided `executor`. + + (Use ProcessPoolExecutors for actual parallelism.) + + `line_length`, `write_back`, and `fast` options are passed to + :func:`format_file_in_place`. + """ tasks = { src: loop.run_in_executor( - executor, format_file_in_place, src, line_length, fast + executor, format_file_in_place, src, line_length, fast, write_back ) for src in sources } @@ -135,44 +197,82 @@ async def schedule_formatting( return report.return_code -def format_file_in_place(src: Path, line_length: int, fast: bool) -> bool: - """Format the file and rewrite if changed. Return True if changed.""" +def format_file_in_place( + src: Path, line_length: int, fast: bool, write_back: bool = False +) -> bool: + """Format file under `src` path. Return True if changed. + + If `write_back` is True, write reformatted code back to stdout. + `line_length` and `fast` options are passed to :func:`format_file_contents`. + """ + with tokenize.open(src) as src_buffer: + src_contents = src_buffer.read() try: - contents, encoding = format_file(src, line_length=line_length, fast=fast) + contents = format_file_contents( + src_contents, line_length=line_length, fast=fast + ) except NothingChanged: return False - with open(src, "w", encoding=encoding) as f: - f.write(contents) + if write_back: + with open(src, "w", encoding=src_buffer.encoding) as f: + f.write(contents) return True -def format_file( - src: Path, line_length: int, fast: bool -) -> Tuple[FileContent, Encoding]: - """Reformats a file and returns its contents and encoding.""" - with tokenize.open(src) as src_buffer: - src_contents = src_buffer.read() +def format_stdin_to_stdout( + line_length: int, fast: bool, write_back: bool = False +) -> bool: + """Format file on stdin. Return True if changed. + + If `write_back` is True, write reformatted code back to stdout. + `line_length` and `fast` arguments are passed to :func:`format_file_contents`. + """ + contents = sys.stdin.read() + try: + contents = format_file_contents(contents, line_length=line_length, fast=fast) + return True + + except NothingChanged: + return False + + finally: + if write_back: + sys.stdout.write(contents) + + +def format_file_contents( + src_contents: str, line_length: int, fast: bool +) -> FileContent: + """Reformats a file and returns its contents and encoding. + + If `fast` is False, additionally confirm that the reformatted code is + valid by calling :func:`assert_equivalent` and :func:`assert_stable` on it. + `line_length` is passed to :func:`format_str`. + """ if src_contents.strip() == '': - raise NothingChanged(src) + raise NothingChanged dst_contents = format_str(src_contents, line_length=line_length) if src_contents == dst_contents: - raise NothingChanged(src) + raise NothingChanged if not fast: assert_equivalent(src_contents, dst_contents) assert_stable(src_contents, dst_contents, line_length=line_length) - return dst_contents, src_buffer.encoding + return dst_contents def format_str(src_contents: str, line_length: int) -> FileContent: - """Reformats a string and returns new contents.""" + """Reformats a string and returns new contents. + + `line_length` determines how many characters per line are allowed. + """ src_node = lib2to3_parse(src_contents) dst_contents = "" - comments: List[Line] = [] lines = LineGenerator() elt = EmptyLineTracker() + py36 = is_python36(src_node) empty_line = Line() after = 0 for current_line in lines.visit(src_node): @@ -181,36 +281,41 @@ def format_str(src_contents: str, line_length: int) -> FileContent: before, after = elt.maybe_empty_lines(current_line) for _ in range(before): dst_contents += str(empty_line) - if not current_line.is_comment: - for comment in comments: - dst_contents += str(comment) - comments = [] - for line in split_line(current_line, line_length=line_length): - dst_contents += str(line) - else: - comments.append(current_line) - for comment in comments: - dst_contents += str(comment) + for line in split_line(current_line, line_length=line_length, py36=py36): + dst_contents += str(line) return dst_contents +GRAMMARS = [ + pygram.python_grammar_no_print_statement_no_exec_statement, + pygram.python_grammar_no_print_statement, + pygram.python_grammar_no_exec_statement, + pygram.python_grammar, +] + + def lib2to3_parse(src_txt: str) -> Node: """Given a string with source, return the lib2to3 Node.""" grammar = pygram.python_grammar_no_print_statement - drv = driver.Driver(grammar, pytree.convert) if src_txt[-1] != '\n': nl = '\r\n' if '\r\n' in src_txt[:1024] else '\n' src_txt += nl - try: - result = drv.parse_string(src_txt, True) - except ParseError as pe: - lineno, column = pe.context[1] - lines = src_txt.splitlines() + for grammar in GRAMMARS: + drv = driver.Driver(grammar, pytree.convert) try: - faulty_line = lines[lineno - 1] - except IndexError: - faulty_line = "" - raise ValueError(f"Cannot parse: {lineno}:{column}: {faulty_line}") from None + result = drv.parse_string(src_txt, True) + break + + except ParseError as pe: + lineno, column = pe.context[1] + lines = src_txt.splitlines() + try: + faulty_line = lines[lineno - 1] + except IndexError: + faulty_line = "" + exc = ValueError(f"Cannot parse: {lineno}:{column}: {faulty_line}") + else: + raise exc from None if isinstance(result, Leaf): result = Node(syms.file_input, [result]) @@ -227,9 +332,18 @@ T = TypeVar('T') class Visitor(Generic[T]): - """Basic lib2to3 visitor that yields things on visiting.""" + """Basic lib2to3 visitor that yields things of type `T` on `visit()`.""" def visit(self, node: LN) -> Iterator[T]: + """Main method to visit `node` and its children. + + It tries to find a `visit_*()` method for the given `node.type`, like + `visit_simple_stmt` for Node objects or `visit_INDENT` for Leaf objects. + If no dedicated `visit_*()` method is found, chooses `visit_default()` + instead. + + Then yields objects of type `T` from the selected visitor. + """ if node.type < 256: name = token.tok_name[node.type] else: @@ -237,6 +351,7 @@ class Visitor(Generic[T]): yield from getattr(self, f'visit_{name}', self.visit_default)(node) def visit_default(self, node: LN) -> Iterator[T]: + """Default `visit_*()` implementation. Recurses to children of `node`.""" if isinstance(node, Node): for child in node.children: yield from self.visit(child) @@ -244,7 +359,7 @@ class Visitor(Generic[T]): @dataclass class DebugVisitor(Visitor[T]): - tree_depth: int = attrib(default=0) + tree_depth: int = 0 def visit_default(self, node: LN) -> Iterator[T]: indent = ' ' * (2 * self.tree_depth) @@ -266,6 +381,15 @@ class DebugVisitor(Visitor[T]): out(f' {node.prefix!r}', fg='green', bold=False, nl=False) out(f' {node.value!r}', fg='blue', bold=False) + @classmethod + def show(cls, code: str) -> None: + """Pretty-prints a given string of `code`. + + Convenience method for debugging. + """ + v: DebugVisitor[None] = DebugVisitor() + list(v.visit(lib2to3_parse(code))) + KEYWORDS = set(keyword.kwlist) WHITESPACE = {token.DEDENT, token.INDENT, token.NEWLINE} @@ -299,6 +423,7 @@ MATH_OPERATORS = { token.AMPER, token.PERCENT, token.CIRCUMFLEX, + token.TILDE, token.LEFTSHIFT, token.RIGHTSHIFT, token.DOUBLESTAR, @@ -314,20 +439,36 @@ MATH_PRIORITY = 1 @dataclass class BracketTracker: - depth: int = attrib(default=0) - bracket_match: Dict[Tuple[Depth, NodeType], Leaf] = attrib(default=Factory(dict)) - delimiters: Dict[LeafID, Priority] = attrib(default=Factory(dict)) - previous: Optional[Leaf] = attrib(default=None) + """Keeps track of brackets on a line.""" + + depth: int = 0 + bracket_match: Dict[Tuple[Depth, NodeType], Leaf] = Factory(dict) + delimiters: Dict[LeafID, Priority] = Factory(dict) + previous: Optional[Leaf] = None def mark(self, leaf: Leaf) -> None: + """Marks `leaf` with bracket-related metadata. Keeps track of delimiters. + + All leaves receive an int `bracket_depth` field that stores how deep + within brackets a given leaf is. 0 means there are no enclosing brackets + that started on this line. + + If a leaf is itself a closing bracket, it receives an `opening_bracket` + field that it forms a pair with. This is a one-directional link to + avoid reference cycles. + + If a leaf is a delimiter (a token on which Black can split the line if + needed) and it's on depth 0, its `id()` is stored in the tracker's + `delimiters` field. + """ if leaf.type == token.COMMENT: return if leaf.type in CLOSING_BRACKETS: self.depth -= 1 opening_bracket = self.bracket_match.pop((self.depth, leaf.type)) - leaf.opening_bracket = opening_bracket # type: ignore - leaf.bracket_depth = self.depth # type: ignore + leaf.opening_bracket = opening_bracket + leaf.bracket_depth = self.depth if self.depth == 0: delim = is_delimiter(leaf) if delim: @@ -336,19 +477,25 @@ class BracketTracker: if leaf.type == token.STRING and self.previous.type == token.STRING: self.delimiters[id(self.previous)] = STRING_PRIORITY elif ( - leaf.type == token.NAME and - leaf.value == 'for' and - leaf.parent and - leaf.parent.type in {syms.comp_for, syms.old_comp_for} + leaf.type == token.NAME + and leaf.value == 'for' + and leaf.parent + and leaf.parent.type in {syms.comp_for, syms.old_comp_for} ): self.delimiters[id(self.previous)] = COMPREHENSION_PRIORITY elif ( - leaf.type == token.NAME and - leaf.value == 'if' and - leaf.parent and - leaf.parent.type in {syms.comp_if, syms.old_comp_if} + leaf.type == token.NAME + and leaf.value == 'if' + and leaf.parent + and leaf.parent.type in {syms.comp_if, syms.old_comp_if} ): self.delimiters[id(self.previous)] = COMPREHENSION_PRIORITY + elif ( + leaf.type == token.NAME + and leaf.value in LOGIC_OPERATORS + and leaf.parent + ): + self.delimiters[id(self.previous)] = LOGIC_PRIORITY if leaf.type in OPENING_BRACKETS: self.bracket_match[self.depth, BRACKET[leaf.type]] = leaf self.depth += 1 @@ -358,7 +505,7 @@ class BracketTracker: """Returns True if there is an yet unmatched open bracket on the line.""" return bool(self.bracket_match) - def max_priority(self, exclude: Iterable[LeafID] = ()) -> int: + def max_delimiter_priority(self, exclude: Iterable[LeafID] = ()) -> int: """Returns the highest priority of a delimiter found on the line. Values are consistent with what `is_delimiter()` returns. @@ -368,13 +515,26 @@ class BracketTracker: @dataclass class Line: - depth: int = attrib(default=0) - leaves: List[Leaf] = attrib(default=Factory(list)) - comments: Dict[LeafID, Leaf] = attrib(default=Factory(dict)) - bracket_tracker: BracketTracker = attrib(default=Factory(BracketTracker)) - inside_brackets: bool = attrib(default=False) + """Holds leaves and comments. Can be printed with `str(line)`.""" + + depth: int = 0 + leaves: List[Leaf] = Factory(list) + comments: Dict[LeafID, Leaf] = Factory(dict) + bracket_tracker: BracketTracker = Factory(BracketTracker) + inside_brackets: bool = False + has_for: bool = False + _for_loop_variable: bool = False def append(self, leaf: Leaf, preformatted: bool = False) -> None: + """Add a new `leaf` to the end of the line. + + Unless `preformatted` is True, the `leaf` will receive a new consistent + whitespace prefix and metadata applied by :class:`BracketTracker`. + Trailing commas are maybe removed, unpacked for loop variables are + demoted from being delimiters. + + Inline comments are put aside. + """ has_value = leaf.value.strip() if not has_value: return @@ -384,8 +544,10 @@ class Line: # imports, for which we only preserve newlines. leaf.prefix += whitespace(leaf) if self.inside_brackets or not preformatted: + self.maybe_decrement_after_for_loop_variable(leaf) self.bracket_tracker.mark(leaf) self.maybe_remove_trailing_comma(leaf) + self.maybe_increment_for_loop_variable(leaf) if self.maybe_adapt_standalone_comment(leaf): return @@ -394,27 +556,31 @@ class Line: @property def is_comment(self) -> bool: + """Is this line a standalone comment?""" return bool(self) and self.leaves[0].type == STANDALONE_COMMENT @property def is_decorator(self) -> bool: + """Is this line a decorator?""" return bool(self) and self.leaves[0].type == token.AT @property def is_import(self) -> bool: + """Is this an import line?""" return bool(self) and is_import(self.leaves[0]) @property def is_class(self) -> bool: + """Is this a class definition?""" return ( - bool(self) and - self.leaves[0].type == token.NAME and - self.leaves[0].value == 'class' + bool(self) + and self.leaves[0].type == token.NAME + and self.leaves[0].value == 'class' ) @property def is_def(self) -> bool: - """Also returns True for async defs.""" + """Is this a function definition? (Also returns True for async defs.)""" try: first_leaf = self.leaves[0] except IndexError: @@ -425,50 +591,61 @@ class Line: except IndexError: second_leaf = None return ( - (first_leaf.type == token.NAME and first_leaf.value == 'def') or - ( - first_leaf.type == token.NAME and - first_leaf.value == 'async' and - second_leaf is not None and - second_leaf.type == token.NAME and - second_leaf.value == 'def' + (first_leaf.type == token.NAME and first_leaf.value == 'def') + or ( + first_leaf.type == token.ASYNC + and second_leaf is not None + and second_leaf.type == token.NAME + and second_leaf.value == 'def' ) ) @property def is_flow_control(self) -> bool: + """Is this a flow control statement? + + Those are `return`, `raise`, `break`, and `continue`. + """ return ( - bool(self) and - self.leaves[0].type == token.NAME and - self.leaves[0].value in FLOW_CONTROL + bool(self) + and self.leaves[0].type == token.NAME + and self.leaves[0].value in FLOW_CONTROL ) @property def is_yield(self) -> bool: + """Is this a yield statement?""" return ( - bool(self) and - self.leaves[0].type == token.NAME and - self.leaves[0].value == 'yield' + bool(self) + and self.leaves[0].type == token.NAME + and self.leaves[0].value == 'yield' ) def maybe_remove_trailing_comma(self, closing: Leaf) -> bool: + """Remove trailing comma if there is one and it's safe.""" if not ( - self.leaves and - self.leaves[-1].type == token.COMMA and - closing.type in CLOSING_BRACKETS + self.leaves + and self.leaves[-1].type == token.COMMA + and closing.type in CLOSING_BRACKETS ): return False - if closing.type == token.RSQB or closing.type == token.RBRACE: + if closing.type == token.RBRACE: self.leaves.pop() return True + if closing.type == token.RSQB: + comma = self.leaves[-1] + if comma.parent and comma.parent.type == syms.listmaker: + self.leaves.pop() + return True + # For parens let's check if it's safe to remove the comma. If the # trailing one is the only one, we might mistakenly change a tuple # into a different type by removing the comma. - depth = closing.bracket_depth + 1 # type: ignore + depth = closing.bracket_depth + 1 commas = 0 - opening = closing.opening_bracket # type: ignore + opening = closing.opening_bracket for _opening_index, leaf in enumerate(self.leaves): if leaf is opening: break @@ -480,15 +657,42 @@ class Line: if leaf is closing: break - bracket_depth = leaf.bracket_depth # type: ignore + bracket_depth = leaf.bracket_depth if bracket_depth == depth and leaf.type == token.COMMA: commas += 1 + if leaf.parent and leaf.parent.type == syms.arglist: + commas += 1 + break + if commas > 1: self.leaves.pop() return True return False + def maybe_increment_for_loop_variable(self, leaf: Leaf) -> bool: + """In a for loop, or comprehension, the variables are often unpacks. + + To avoid splitting on the comma in this situation, we will increase + the depth of tokens between `for` and `in`. + """ + if leaf.type == token.NAME and leaf.value == 'for': + self.has_for = True + self.bracket_tracker.depth += 1 + self._for_loop_variable = True + return True + + return False + + def maybe_decrement_after_for_loop_variable(self, leaf: Leaf) -> bool: + """See `maybe_increment_for_loop_variable` above for explanation.""" + if self._for_loop_variable and leaf.type == token.NAME and leaf.value == 'in': + self.bracket_tracker.depth -= 1 + self._for_loop_variable = False + return True + + return False + def maybe_adapt_standalone_comment(self, comment: Leaf) -> bool: """Hack a standalone comment to act as a trailing comment for line splitting. @@ -499,8 +703,8 @@ class Line: appended will appear "too long" when splitting. """ if not ( - comment.type == STANDALONE_COMMENT and - self.bracket_tracker.any_open_brackets() + comment.type == STANDALONE_COMMENT + and self.bracket_tracker.any_open_brackets() ): return False @@ -509,6 +713,7 @@ class Line: return self.append_comment(comment) def append_comment(self, comment: Leaf) -> bool: + """Add an inline comment to the line.""" if comment.type != token.COMMENT: return False @@ -527,6 +732,7 @@ class Line: return True def last_non_delimiter(self) -> Leaf: + """Returns the last non-delimiter on the line. Raises LookupError otherwise.""" for i in range(len(self.leaves)): last = self.leaves[-i - 1] if not is_delimiter(last): @@ -535,6 +741,7 @@ class Line: raise LookupError("No non-delimiters found") def __str__(self) -> str: + """Render the line.""" if not self: return '\n' @@ -549,19 +756,75 @@ class Line: return res + '\n' def __bool__(self) -> bool: + """Returns True if the line has leaves or comments.""" return bool(self.leaves or self.comments) +class UnformattedLines(Line): + """Just like :class:`Line` but stores lines which aren't reformatted.""" + + def append(self, leaf: Leaf, preformatted: bool = True) -> None: + """Just add a new `leaf` to the end of the lines. + + The `preformatted` argument is ignored. + + Keeps track of indentation `depth`, which is useful when the user + says `# fmt: on`. Otherwise, doesn't do anything with the `leaf`. + """ + try: + list(generate_comments(leaf)) + except FormatOn as f_on: + self.leaves.append(f_on.leaf_from_consumed(leaf)) + raise + + self.leaves.append(leaf) + if leaf.type == token.INDENT: + self.depth += 1 + elif leaf.type == token.DEDENT: + self.depth -= 1 + + def append_comment(self, comment: Leaf) -> bool: + """Not implemented in this class.""" + raise NotImplementedError("Unformatted lines don't store comments separately.") + + def maybe_remove_trailing_comma(self, closing: Leaf) -> bool: + """Does nothing and returns False.""" + return False + + def maybe_increment_for_loop_variable(self, leaf: Leaf) -> bool: + """Does nothing and returns False.""" + return False + + def maybe_adapt_standalone_comment(self, comment: Leaf) -> bool: + """Does nothing and returns False.""" + return False + + def __str__(self) -> str: + """Renders unformatted lines from leaves which were added with `append()`. + + `depth` is not used for indentation in this case. + """ + if not self: + return '\n' + + res = '' + for leaf in self.leaves: + res += str(leaf) + return res + + @dataclass class EmptyLineTracker: """Provides a stateful method that returns the number of potential extra empty lines needed before and after the currently processed line. - Note: this tracker works on lines that haven't been split yet. + Note: this tracker works on lines that haven't been split yet. It assumes + the prefix of the first leaf consists of optional newlines. Those newlines + are consumed by `maybe_empty_lines()` and included in the computation. """ - previous_line: Optional[Line] = attrib(default=None) - previous_after: int = attrib(default=0) - previous_defs: List[int] = attrib(default=Factory(list)) + previous_line: Optional[Line] = None + previous_after: int = 0 + previous_defs: List[int] = Factory(list) def maybe_empty_lines(self, current_line: Line) -> Tuple[int, int]: """Returns the number of extra empty lines before and after the `current_line`. @@ -570,17 +833,31 @@ class EmptyLineTracker: (two on module-level), as well as providing an extra empty line after flow control keywords to make them more prominent. """ + if isinstance(current_line, UnformattedLines): + return 0, 0 + before, after = self._maybe_empty_lines(current_line) + before -= self.previous_after self.previous_after = after self.previous_line = current_line return before, after def _maybe_empty_lines(self, current_line: Line) -> Tuple[int, int]: - before = 0 + max_allowed = 1 + if current_line.depth == 0: + max_allowed = 2 + if current_line.leaves: + # Consume the first leaf's extra newlines. + first_leaf = current_line.leaves[0] + before = first_leaf.prefix.count('\n') + before = min(before, max_allowed) + first_leaf.prefix = '' + else: + before = 0 depth = current_line.depth while self.previous_defs and self.previous_defs[-1] >= depth: self.previous_defs.pop() - before = (1 if depth else 2) - self.previous_after + before = 1 if depth else 2 is_decorator = current_line.is_decorator if is_decorator or current_line.is_def or current_line.is_class: if not is_decorator: @@ -596,24 +873,23 @@ class EmptyLineTracker: newlines = 2 if current_line.depth: newlines -= 1 - newlines -= self.previous_after return newlines, 0 if current_line.is_flow_control: return before, 1 if ( - self.previous_line and - self.previous_line.is_import and - not current_line.is_import and - depth == self.previous_line.depth + self.previous_line + and self.previous_line.is_import + and not current_line.is_import + and depth == self.previous_line.depth ): return (before or 1), 0 if ( - self.previous_line and - self.previous_line.is_yield and - (not current_line.is_yield or depth != self.previous_line.depth) + self.previous_line + and self.previous_line.is_yield + and (not current_line.is_yield or depth != self.previous_line.depth) ): return (before or 1), 0 @@ -627,10 +903,9 @@ class LineGenerator(Visitor[Line]): Note: destroys the tree it's visiting by mutating prefixes of its leaves in ways that will no longer stringify to valid Python code on the tree. """ - current_line: Line = attrib(default=Factory(Line)) - standalone_comments: List[Leaf] = attrib(default=Factory(list)) + current_line: Line = Factory(Line) - def line(self, indent: int = 0) -> Iterator[Line]: + def line(self, indent: int = 0, type: Type[Line] = Line) -> Iterator[Line]: """Generate a line. If the line is empty, only emit if it makes sense. @@ -639,59 +914,82 @@ class LineGenerator(Visitor[Line]): If any lines were generated, set up a new current_line. """ if not self.current_line: - self.current_line.depth += indent + if self.current_line.__class__ == type: + self.current_line.depth += indent + else: + self.current_line = type(depth=self.current_line.depth + indent) return # Line is empty, don't emit. Creating a new one unnecessary. complete_line = self.current_line - self.current_line = Line(depth=complete_line.depth + indent) + self.current_line = type(depth=complete_line.depth + indent) yield complete_line + def visit(self, node: LN) -> Iterator[Line]: + """Main method to start the visit process. Yields :class:`Line` objects.""" + if isinstance(self.current_line, UnformattedLines): + # File contained `# fmt: off` + yield from self.visit_unformatted(node) + + else: + yield from super().visit(node) + def visit_default(self, node: LN) -> Iterator[Line]: + """Default `visit_*()` implementation. Recurses to children of `node`.""" if isinstance(node, Leaf): - for comment in generate_comments(node): - if self.current_line.bracket_tracker.any_open_brackets(): - # any comment within brackets is subject to splitting - self.current_line.append(comment) - elif comment.type == token.COMMENT: - # regular trailing comment - self.current_line.append(comment) - yield from self.line() - - else: - # regular standalone comment, to be processed later (see - # docstring in `generate_comments()` - self.standalone_comments.append(comment) - normalize_prefix(node) - if node.type not in WHITESPACE: - for comment in self.standalone_comments: - yield from self.line() - - self.current_line.append(comment) - yield from self.line() - - self.standalone_comments = [] - self.current_line.append(node) + any_open_brackets = self.current_line.bracket_tracker.any_open_brackets() + try: + for comment in generate_comments(node): + if any_open_brackets: + # any comment within brackets is subject to splitting + self.current_line.append(comment) + elif comment.type == token.COMMENT: + # regular trailing comment + self.current_line.append(comment) + yield from self.line() + + else: + # regular standalone comment + yield from self.line() + + self.current_line.append(comment) + yield from self.line() + + except FormatOff as f_off: + f_off.trim_prefix(node) + yield from self.line(type=UnformattedLines) + yield from self.visit(node) + + except FormatOn as f_on: + # This only happens here if somebody says "fmt: on" multiple + # times in a row. + f_on.trim_prefix(node) + yield from self.visit_default(node) + + else: + normalize_prefix(node, inside_brackets=any_open_brackets) + if node.type not in WHITESPACE: + self.current_line.append(node) yield from super().visit_default(node) - def visit_suite(self, node: Node) -> Iterator[Line]: - """Body of a statement after a colon.""" - children = iter(node.children) - # Process newline before indenting. It might contain an inline - # comment that should go right after the colon. - newline = next(children) - yield from self.visit(newline) + def visit_INDENT(self, node: Node) -> Iterator[Line]: + """Increases indentation level, maybe yields a line.""" + # In blib2to3 INDENT never holds comments. yield from self.line(+1) + yield from self.visit_default(node) - for child in children: - yield from self.visit(child) - + def visit_DEDENT(self, node: Node) -> Iterator[Line]: + """Decreases indentation level, maybe yields a line.""" + # DEDENT has no value. Additionally, in blib2to3 it never holds comments. yield from self.line(-1) def visit_stmt(self, node: Node, keywords: Set[str]) -> Iterator[Line]: - """Visit a statement. + """Visits a statement. - The relevant Python language keywords for this statement are NAME leaves - within it. + This implementation is shared for `if`, `while`, `for`, `try`, `except`, + `def`, `with`, and `class`. + + The relevant Python language `keywords` for a given statement will be NAME + leaves within it. This methods puts those on a separate line. """ for child in node.children: if child.type == token.NAME and child.value in keywords: # type: ignore @@ -700,7 +998,7 @@ class LineGenerator(Visitor[Line]): yield from self.visit(child) def visit_simple_stmt(self, node: Node) -> Iterator[Line]: - """A statement without nested statements.""" + """Visits a statement without nested statements.""" is_suite_like = node.parent and node.parent.type in STATEMENT if is_suite_like: yield from self.line(+1) @@ -712,13 +1010,14 @@ class LineGenerator(Visitor[Line]): yield from self.visit_default(node) def visit_async_stmt(self, node: Node) -> Iterator[Line]: + """Visits `async def`, `async for`, `async with`.""" yield from self.line() children = iter(node.children) for child in children: yield from self.visit(child) - if child.type == token.NAME and child.value == 'async': # type: ignore + if child.type == token.ASYNC: break internal_stmt = next(children) @@ -726,17 +1025,40 @@ class LineGenerator(Visitor[Line]): yield from self.visit(child) def visit_decorators(self, node: Node) -> Iterator[Line]: + """Visits decorators.""" for child in node.children: yield from self.line() yield from self.visit(child) def visit_SEMI(self, leaf: Leaf) -> Iterator[Line]: + """Semicolons are always removed. + + Statements between them are put on separate lines. + """ yield from self.line() def visit_ENDMARKER(self, leaf: Leaf) -> Iterator[Line]: + """End of file. + + Process outstanding comments and end with a newline. + """ yield from self.visit_default(leaf) yield from self.line() + def visit_unformatted(self, node: LN) -> Iterator[Line]: + """Used when file contained a `# fmt: off`.""" + if isinstance(node, Node): + for child in node.children: + yield from self.visit(child) + + else: + try: + self.current_line.append(node) + except FormatOn as f_on: + f_on.trim_prefix(node) + yield from self.line() + yield from self.visit(node) + def __attrs_post_init__(self) -> None: """You are in a twisty little maze of passages.""" v = self.visit_stmt @@ -756,9 +1078,10 @@ BRACKET = {token.LPAR: token.RPAR, token.LSQB: token.RSQB, token.LBRACE: token.R OPENING_BRACKETS = set(BRACKET.keys()) CLOSING_BRACKETS = set(BRACKET.values()) BRACKETS = OPENING_BRACKETS | CLOSING_BRACKETS +ALWAYS_NO_SPACE = CLOSING_BRACKETS | {token.COMMA, STANDALONE_COMMENT} -def whitespace(leaf: Leaf) -> str: +def whitespace(leaf: Leaf) -> str: # noqa C901 """Return whitespace prefix if needed for the given `leaf`.""" NO = '' SPACE = ' ' @@ -766,43 +1089,92 @@ def whitespace(leaf: Leaf) -> str: t = leaf.type p = leaf.parent v = leaf.value - if t == token.COLON: + if t in ALWAYS_NO_SPACE: return NO - if t == token.COMMA: - return NO + if t == token.COMMENT: + return DOUBLESPACE - if t == token.RPAR: + assert p is not None, f"INTERNAL ERROR: hand-made leaf without parent: {leaf!r}" + if t == token.COLON and p.type not in {syms.subscript, syms.subscriptlist}: return NO - if t == token.COMMENT: - return DOUBLESPACE + prev = leaf.prev_sibling + if not prev: + prevp = preceding_leaf(p) + if not prevp or prevp.type in OPENING_BRACKETS: + return NO + + if t == token.COLON: + return SPACE if prevp.type == token.COMMA else NO + + if prevp.type == token.EQUAL: + if prevp.parent: + if prevp.parent.type in { + syms.arglist, syms.argument, syms.parameters, syms.varargslist + }: + return NO + + elif prevp.parent.type == syms.typedargslist: + # A bit hacky: if the equal sign has whitespace, it means we + # previously found it's a typed argument. So, we're using + # that, too. + return prevp.prefix + + elif prevp.type == token.DOUBLESTAR: + if prevp.parent and prevp.parent.type in { + syms.arglist, + syms.argument, + syms.dictsetmaker, + syms.parameters, + syms.typedargslist, + syms.varargslist, + }: + return NO + + elif prevp.type == token.COLON: + if prevp.parent and prevp.parent.type in {syms.subscript, syms.sliceop}: + return NO + + elif ( + prevp.parent + and prevp.parent.type in {syms.factor, syms.star_expr} + and prevp.type in MATH_OPERATORS + ): + return NO - if t == STANDALONE_COMMENT: + elif ( + prevp.type == token.RIGHTSHIFT + and prevp.parent + and prevp.parent.type == syms.shift_expr + and prevp.prev_sibling + and prevp.prev_sibling.type == token.NAME + and prevp.prev_sibling.value == 'print' # type: ignore + ): + # Python 2 print chevron + return NO + + elif prev.type in OPENING_BRACKETS: return NO - assert p is not None, f"INTERNAL ERROR: hand-made leaf without parent: {leaf!r}" if p.type in {syms.parameters, syms.arglist}: # untyped function signatures or calls if t == token.RPAR: return NO - prev = leaf.prev_sibling if not prev or prev.type != token.COMMA: return NO - if p.type == syms.varargslist: + elif p.type == syms.varargslist: # lambdas if t == token.RPAR: return NO - prev = leaf.prev_sibling if prev and prev.type != token.COMMA: return NO elif p.type == syms.typedargslist: # typed function signatures - prev = leaf.prev_sibling if not prev: return NO @@ -820,7 +1192,6 @@ def whitespace(leaf: Leaf) -> str: elif p.type == syms.tname: # type names - prev = leaf.prev_sibling if not prev: prevp = preceding_leaf(p) if not prevp or prevp.type != token.COMMA: @@ -831,7 +1202,6 @@ def whitespace(leaf: Leaf) -> str: if t == token.LPAR or t == token.RPAR: return NO - prev = leaf.prev_sibling if not prev: if t == token.DOT: prevp = preceding_leaf(p) @@ -849,7 +1219,6 @@ def whitespace(leaf: Leaf) -> str: if t == token.EQUAL: return NO - prev = leaf.prev_sibling if not prev: prevp = preceding_leaf(p) if not prevp or prevp.type == token.LPAR: @@ -863,7 +1232,6 @@ def whitespace(leaf: Leaf) -> str: return NO elif p.type == syms.dotted_name: - prev = leaf.prev_sibling if prev: return NO @@ -875,101 +1243,45 @@ def whitespace(leaf: Leaf) -> str: if t == token.LPAR: return NO - prev = leaf.prev_sibling if prev and prev.type == token.LPAR: return NO elif p.type == syms.subscript: # indexing - if t == token.COLON: - return NO - - prev = leaf.prev_sibling - if not prev or prev.type == token.COLON: - return NO - - elif p.type in { - syms.test, - syms.not_test, - syms.xor_expr, - syms.or_test, - syms.and_test, - syms.arith_expr, - syms.shift_expr, - syms.yield_expr, - syms.term, - syms.power, - syms.comparison, - }: - # various arithmetic and logic expressions - prev = leaf.prev_sibling if not prev: - prevp = preceding_leaf(p) - if not prevp or prevp.type in OPENING_BRACKETS: - return NO - - if prevp.type == token.EQUAL: - if prevp.parent and prevp.parent.type in { - syms.varargslist, syms.parameters, syms.arglist, syms.argument - }: - return NO - - return SPACE + assert p.parent is not None, "subscripts are always parented" + if p.parent.type == syms.subscriptlist: + return SPACE - elif p.type == syms.atom: - if t in CLOSING_BRACKETS: return NO - prev = leaf.prev_sibling - if not prev: - prevp = preceding_leaf(p) - if not prevp: - return NO - - if prevp.type in OPENING_BRACKETS: - return NO - - if prevp.type == token.EQUAL: - if prevp.parent and prevp.parent.type in { - syms.varargslist, syms.parameters, syms.arglist, syms.argument - }: - return NO - - if prevp.type == token.DOUBLESTAR: - if prevp.parent and prevp.parent.type in { - syms.varargslist, syms.parameters, syms.arglist, syms.dictsetmaker - }: - return NO - - elif prev.type in OPENING_BRACKETS: + else: return NO - elif t == token.DOT: + elif p.type == syms.atom: + if prev and t == token.DOT: # dots, but not the first one. return NO elif ( - p.type == syms.listmaker or - p.type == syms.testlist_gexp or - p.type == syms.subscriptlist + p.type == syms.listmaker + or p.type == syms.testlist_gexp + or p.type == syms.subscriptlist ): # list interior, including unpacking - prev = leaf.prev_sibling if not prev: return NO elif p.type == syms.dictsetmaker: # dict and set interior, including unpacking - prev = leaf.prev_sibling if not prev: return NO if prev.type == token.DOUBLESTAR: return NO - elif p.type == syms.factor or p.type == syms.star_expr: + elif p.type in {syms.factor, syms.star_expr}: # unary ops - prev = leaf.prev_sibling if not prev: prevp = preceding_leaf(p) if not prevp or prevp.type in OPENING_BRACKETS: @@ -990,7 +1302,6 @@ def whitespace(leaf: Leaf) -> str: elif p.type == syms.import_from: if t == token.DOT: - prev = leaf.prev_sibling if prev and prev.type == token.DOT: return NO @@ -998,7 +1309,6 @@ def whitespace(leaf: Leaf) -> str: if v == 'import': return SPACE - prev = leaf.prev_sibling if prev and prev.type == token.DOT: return NO @@ -1034,16 +1344,13 @@ def is_delimiter(leaf: Leaf) -> int: if leaf.type == token.COMMA: return COMMA_PRIORITY - if leaf.type == token.NAME and leaf.value in LOGIC_OPERATORS: - return LOGIC_PRIORITY - if leaf.type in COMPARATORS: return COMPARATOR_PRIORITY if ( - leaf.type in MATH_OPERATORS and - leaf.parent and - leaf.parent.type not in {syms.factor, syms.star_expr} + leaf.type in MATH_OPERATORS + and leaf.parent + and leaf.parent.type not in {syms.factor, syms.star_expr} ): return MATH_PRIORITY @@ -1069,40 +1376,75 @@ def generate_comments(leaf: Leaf) -> Iterator[Leaf]: Inline comments are emitted as regular token.COMMENT leaves. Standalone are emitted with a fake STANDALONE_COMMENT token identifier. """ - if not leaf.prefix: - return - - if '#' not in leaf.prefix: + p = leaf.prefix + if not p: return - before_comment, content = leaf.prefix.split('#', 1) - content = content.rstrip() - if content and (content[0] not in {' ', '!', '#'}): - content = ' ' + content - is_standalone_comment = ( - '\n' in before_comment or '\n' in content or leaf.type == token.DEDENT - ) - if not is_standalone_comment: - # simple trailing comment - yield Leaf(token.COMMENT, value='#' + content) + if '#' not in p: return - for line in ('#' + content).split('\n'): + consumed = 0 + nlines = 0 + for index, line in enumerate(p.split('\n')): + consumed += len(line) + 1 # adding the length of the split '\n' line = line.lstrip() + if not line: + nlines += 1 if not line.startswith('#'): continue - yield Leaf(STANDALONE_COMMENT, line) + if index == 0 and leaf.type != token.ENDMARKER: + comment_type = token.COMMENT # simple trailing comment + else: + comment_type = STANDALONE_COMMENT + comment = make_comment(line) + yield Leaf(comment_type, comment, prefix='\n' * nlines) + + if comment in {'# fmt: on', '# yapf: enable'}: + raise FormatOn(consumed) + + if comment in {'# fmt: off', '# yapf: disable'}: + raise FormatOff(consumed) + + nlines = 0 + +def make_comment(content: str) -> str: + """Returns a consistently formatted comment from the given `content` string. -def split_line(line: Line, line_length: int, inner: bool = False) -> Iterator[Line]: + All comments (except for "##", "#!", "#:") should have a single space between + the hash sign and the content. + + If `content` didn't start with a hash sign, one is provided. + """ + content = content.rstrip() + if not content: + return '#' + + if content[0] == '#': + content = content[1:] + if content and content[0] not in ' !:#': + content = ' ' + content + return '#' + content + + +def split_line( + line: Line, line_length: int, inner: bool = False, py36: bool = False +) -> Iterator[Line]: """Splits a `line` into potentially many lines. They should fit in the allotted `line_length` but might not be able to. `inner` signifies that there were a pair of brackets somewhere around the current `line`, possibly transitively. This means we can fallback to splitting by delimiters if the LHS/RHS don't yield any results. + + If `py36` is True, splitting may generate syntax that is only compatible + with Python 3.6 and later. """ + if isinstance(line, UnformattedLines): + yield line + return + line_str = str(line).strip('\n') if len(line_str) <= line_length and '\n' not in line_str: yield line @@ -1124,11 +1466,13 @@ def split_line(line: Line, line_length: int, inner: bool = False) -> Iterator[Li # split altogether. result: List[Line] = [] try: - for l in split_func(line): + for l in split_func(line, py36=py36): if str(l).strip('\n') == line_str: raise CannotSplit("Split function returned an unchanged result") - result.extend(split_line(l, line_length=line_length, inner=True)) + result.extend( + split_line(l, line_length=line_length, inner=True, py36=py36) + ) except CannotSplit as cs: continue @@ -1140,8 +1484,8 @@ def split_line(line: Line, line_length: int, inner: bool = False) -> Iterator[Li yield line -def left_hand_split(line: Line) -> Iterator[Line]: - """Split line into many lines, starting with the first matching bracket pair. +def left_hand_split(line: Line, py36: bool = False) -> Iterator[Line]: + """Splits line into many lines, starting with the first matching bracket pair. Note: this usually looks weird, only use this for function definitions. Prefer RHS otherwise. @@ -1156,11 +1500,11 @@ def left_hand_split(line: Line) -> Iterator[Line]: matching_bracket = None for leaf in line.leaves: if ( - current_leaves is body_leaves and - leaf.type in CLOSING_BRACKETS and - leaf.opening_bracket is matching_bracket # type: ignore + current_leaves is body_leaves + and leaf.type in CLOSING_BRACKETS + and leaf.opening_bracket is matching_bracket ): - current_leaves = tail_leaves + current_leaves = tail_leaves if body_leaves else head_leaves current_leaves.append(leaf) if current_leaves is head_leaves: if leaf.type in OPENING_BRACKETS: @@ -1168,7 +1512,7 @@ def left_hand_split(line: Line) -> Iterator[Line]: current_leaves = body_leaves # Since body is a new indent level, remove spurious leading whitespace. if body_leaves: - normalize_prefix(body_leaves[0]) + normalize_prefix(body_leaves[0], inside_brackets=True) # Build the new lines. for result, leaves in ( (head, head_leaves), (body, body_leaves), (tail, tail_leaves) @@ -1178,25 +1522,14 @@ def left_hand_split(line: Line) -> Iterator[Line]: comment_after = line.comments.get(id(leaf)) if comment_after: result.append(comment_after, preformatted=True) - # Check if the split succeeded. - tail_len = len(str(tail)) - if not body: - if tail_len == 0: - raise CannotSplit("Splitting brackets produced the same line") - - elif tail_len < 3: - raise CannotSplit( - f"Splitting brackets on an empty body to save " - f"{tail_len} characters is not worth it" - ) - + bracket_split_succeeded_or_raise(head, body, tail) for result in (head, body, tail): if result: yield result -def right_hand_split(line: Line) -> Iterator[Line]: - """Split line into many lines, starting with the last matching bracket pair.""" +def right_hand_split(line: Line, py36: bool = False) -> Iterator[Line]: + """Splits line into many lines, starting with the last matching bracket pair.""" head = Line(depth=line.depth) body = Line(depth=line.depth + 1, inside_brackets=True) tail = Line(depth=line.depth) @@ -1208,18 +1541,18 @@ def right_hand_split(line: Line) -> Iterator[Line]: for leaf in reversed(line.leaves): if current_leaves is body_leaves: if leaf is opening_bracket: - current_leaves = head_leaves + current_leaves = head_leaves if body_leaves else tail_leaves current_leaves.append(leaf) if current_leaves is tail_leaves: if leaf.type in CLOSING_BRACKETS: - opening_bracket = leaf.opening_bracket # type: ignore + opening_bracket = leaf.opening_bracket current_leaves = body_leaves tail_leaves.reverse() body_leaves.reverse() head_leaves.reverse() # Since body is a new indent level, remove spurious leading whitespace. if body_leaves: - normalize_prefix(body_leaves[0]) + normalize_prefix(body_leaves[0], inside_brackets=True) # Build the new lines. for result, leaves in ( (head, head_leaves), (body, body_leaves), (tail, tail_leaves) @@ -1229,8 +1562,27 @@ def right_hand_split(line: Line) -> Iterator[Line]: comment_after = line.comments.get(id(leaf)) if comment_after: result.append(comment_after, preformatted=True) - # Check if the split succeeded. - tail_len = len(str(tail).strip('\n')) + bracket_split_succeeded_or_raise(head, body, tail) + for result in (head, body, tail): + if result: + yield result + + +def bracket_split_succeeded_or_raise(head: Line, body: Line, tail: Line) -> None: + """Raise :exc:`CannotSplit` if the last left- or right-hand split failed. + + Do nothing otherwise. + + A left- or right-hand split is based on a pair of brackets. Content before + (and including) the opening bracket is left on one line, content inside the + brackets is put on a separate line, and finally content starting with and + following the closing bracket is put on a separate line. + + Those are called `head`, `body`, and `tail`, respectively. If the split + produced the same line (all content in `head`) or ended up with an empty `body` + and the `tail` is just the closing bracket, then it's considered failed. + """ + tail_len = len(str(tail).strip()) if not body: if tail_len == 0: raise CannotSplit("Splitting brackets produced the same line") @@ -1241,15 +1593,13 @@ def right_hand_split(line: Line) -> Iterator[Line]: f"{tail_len} characters is not worth it" ) - for result in (head, body, tail): - if result: - yield result - -def delimiter_split(line: Line) -> Iterator[Line]: - """Split according to delimiters of the highest priority. +def delimiter_split(line: Line, py36: bool = False) -> Iterator[Line]: + """Splits according to delimiters of the highest priority. This kind of split doesn't increase indentation. + If `py36` is True, the split will add trailing commas also in function + signatures that contain `*` and `**`. """ try: last_leaf = line.leaves[-1] @@ -1258,29 +1608,41 @@ def delimiter_split(line: Line) -> Iterator[Line]: delimiters = line.bracket_tracker.delimiters try: - delimiter_priority = line.bracket_tracker.max_priority(exclude={id(last_leaf)}) + delimiter_priority = line.bracket_tracker.max_delimiter_priority( + exclude={id(last_leaf)} + ) except ValueError: raise CannotSplit("No delimiters found") current_line = Line(depth=line.depth, inside_brackets=line.inside_brackets) + lowest_depth = sys.maxsize + trailing_comma_safe = True for leaf in line.leaves: current_line.append(leaf, preformatted=True) comment_after = line.comments.get(id(leaf)) if comment_after: current_line.append(comment_after, preformatted=True) + lowest_depth = min(lowest_depth, leaf.bracket_depth) + if ( + leaf.bracket_depth == lowest_depth + and leaf.type == token.STAR + or leaf.type == token.DOUBLESTAR + ): + trailing_comma_safe = trailing_comma_safe and py36 leaf_priority = delimiters.get(id(leaf)) if leaf_priority == delimiter_priority: - normalize_prefix(current_line.leaves[0]) + normalize_prefix(current_line.leaves[0], inside_brackets=True) yield current_line current_line = Line(depth=line.depth, inside_brackets=line.inside_brackets) if current_line: if ( - delimiter_priority == COMMA_PRIORITY and - current_line.leaves[-1].type != token.COMMA + delimiter_priority == COMMA_PRIORITY + and current_line.leaves[-1].type != token.COMMA + and trailing_comma_safe ): current_line.append(Leaf(token.COMMA, ',')) - normalize_prefix(current_line.leaves[0]) + normalize_prefix(current_line.leaves[0], inside_brackets=True) yield current_line @@ -1290,28 +1652,57 @@ def is_import(leaf: Leaf) -> bool: t = leaf.type v = leaf.value return bool( - t == token.NAME and - ( - (v == 'import' and p and p.type == syms.import_name) or - (v == 'from' and p and p.type == syms.import_from) + t == token.NAME + and ( + (v == 'import' and p and p.type == syms.import_name) + or (v == 'from' and p and p.type == syms.import_from) ) ) -def normalize_prefix(leaf: Leaf) -> None: - """Leave existing extra newlines for imports. Remove everything else.""" - if is_import(leaf): - spl = leaf.prefix.split('#', 1) - nl_count = spl[0].count('\n') - if len(spl) > 1: - # Skip one newline since it was for a standalone comment. - nl_count -= 1 - leaf.prefix = '\n' * nl_count - return +def normalize_prefix(leaf: Leaf, *, inside_brackets: bool) -> None: + """Leaves existing extra newlines if not `inside_brackets`. + + Removes everything else. Note: don't use backslashes for formatting or + you'll lose your voting rights. + """ + if not inside_brackets: + spl = leaf.prefix.split('#') + if '\\' not in spl[0]: + nl_count = spl[-1].count('\n') + if len(spl) > 1: + nl_count -= 1 + leaf.prefix = '\n' * nl_count + return leaf.prefix = '' +def is_python36(node: Node) -> bool: + """Returns True if the current file is using Python 3.6+ features. + + Currently looking for: + - f-strings; and + - trailing commas after * or ** in function signatures. + """ + for n in node.pre_order(): + if n.type == token.STRING: + value_head = n.value[:2] # type: ignore + if value_head in {'f"', 'F"', "f'", "F'", 'rf', 'fr', 'RF', 'FR'}: + return True + + elif ( + n.type == syms.typedargslist + and n.children + and n.children[-1].type == token.COMMA + ): + for ch in n.children: + if ch.type == token.STAR or ch.type == token.DOUBLESTAR: + return True + + return False + + PYTHON_EXTENSIONS = {'.py'} BLACKLISTED_DIRECTORIES = { 'build', 'buck-out', 'dist', '_build', '.git', '.hg', '.mypy_cache', '.tox', '.venv' @@ -1319,6 +1710,9 @@ BLACKLISTED_DIRECTORIES = { def gen_python_files_in_dir(path: Path) -> Iterator[Path]: + """Generates all files under `path` which aren't under BLACKLISTED_DIRECTORIES + and have one of the PYTHON_EXTENSIONS. + """ for child in path.iterdir(): if child.is_dir(): if child.name in BLACKLISTED_DIRECTORIES: @@ -1332,15 +1726,17 @@ def gen_python_files_in_dir(path: Path) -> Iterator[Path]: @dataclass class Report: - """Provides a reformatting counter.""" - change_count: int = attrib(default=0) - same_count: int = attrib(default=0) - failure_count: int = attrib(default=0) + """Provides a reformatting counter. Can be rendered with `str(report)`.""" + check: bool = False + change_count: int = 0 + same_count: int = 0 + failure_count: int = 0 def done(self, src: Path, changed: bool) -> None: """Increment the counter for successful reformatting. Write out a message.""" if changed: - out(f'reformatted {src}') + reformatted = 'would reformat' if self.check else 'reformatted' + out(f'{reformatted} {src}') self.change_count += 1 else: out(f'{src} already well formatted, good job.', bold=False) @@ -1354,28 +1750,42 @@ class Report: @property def return_code(self) -> int: """Which return code should the app use considering the current state.""" - return 1 if self.failure_count else 0 + # According to http://tldp.org/LDP/abs/html/exitcodes.html starting with + # 126 we have special returncodes reserved by the shell. + if self.failure_count: + return 123 + + elif self.change_count and self.check: + return 1 + + return 0 def __str__(self) -> str: """A color report of the current state. Use `click.unstyle` to remove colors. """ + if self.check: + reformatted = "would be reformatted" + unchanged = "would be left unchanged" + failed = "would fail to reformat" + else: + reformatted = "reformatted" + unchanged = "left unchanged" + failed = "failed to reformat" report = [] if self.change_count: s = 's' if self.change_count > 1 else '' report.append( - click.style(f'{self.change_count} file{s} reformatted', bold=True) + click.style(f'{self.change_count} file{s} {reformatted}', bold=True) ) if self.same_count: s = 's' if self.same_count > 1 else '' - report.append(f'{self.same_count} file{s} left unchanged') + report.append(f'{self.same_count} file{s} {unchanged}') if self.failure_count: s = 's' if self.failure_count > 1 else '' report.append( - click.style( - f'{self.failure_count} file{s} failed to reformat', fg='red' - ) + click.style(f'{self.failure_count} file{s} {failed}', fg='red') ) return ', '.join(report) + '.' @@ -1417,7 +1827,12 @@ def assert_equivalent(src: str, dst: str) -> None: try: src_ast = ast.parse(src) except Exception as exc: - raise AssertionError(f"cannot parse source: {exc}") from None + major, minor = sys.version_info[:2] + raise AssertionError( + f"cannot use --safe with this file; failed to parse source file " + f"with Python {major}.{minor}'s builtin AST. Re-run with --fast " + f"or stop using deprecated Python 2 syntax. AST error message: {exc}" + ) try: dst_ast = ast.parse(dst) @@ -1426,7 +1841,7 @@ def assert_equivalent(src: str, dst: str) -> None: raise AssertionError( f"INTERNAL ERROR: Black produced invalid code: {exc}. " f"Please report a bug on https://github.com/ambv/black/issues. " - f"This invalid output might be helpful: {log}", + f"This invalid output might be helpful: {log}" ) from None src_ast_str = '\n'.join(_v(src_ast)) @@ -1437,7 +1852,7 @@ def assert_equivalent(src: str, dst: str) -> None: f"INTERNAL ERROR: Black produced code that is not equivalent to " f"the source. " f"Please report a bug on https://github.com/ambv/black/issues. " - f"This diff might be helpful: {log}", + f"This diff might be helpful: {log}" ) from None @@ -1456,7 +1871,7 @@ def assert_stable(src: str, dst: str, line_length: int) -> None: f"INTERNAL ERROR: Black produced different code on the second pass " f"of the formatter. " f"Please report a bug on https://github.com/ambv/black/issues. " - f"This diff might be helpful: {log}", + f"This diff might be helpful: {log}" ) from None