X-Git-Url: https://git.madduck.net/etc/vim.git/blobdiff_plain/4c480bcac288883040e6fda3c75825a0da54eb66..0677a539370b296399854e427ce7df2955ecfe57:/black.py diff --git a/black.py b/black.py index e2c427c..dc02128 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__ = "18.3a3" DEFAULT_LINE_LENGTH = 88 # types syms = pygram.python_symbols @@ -42,10 +44,38 @@ class NothingChanged(UserWarning): 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 fmt: on/off error. + + 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( '-l', @@ -60,7 +90,7 @@ class CannotSplit(Exception): is_flag=True, help=( "Don't write back the files, just return the status. Return code 0 " - "means nothing changed. Return code 1 means some files were " + "means nothing would change. Return code 1 means some files would be " "reformatted. Return code 123 means there was an internal error." ), ) @@ -73,7 +103,9 @@ 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( @@ -88,17 +120,24 @@ def main( 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, write_back=not check - ) + 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)) @@ -155,43 +194,62 @@ def format_file_in_place( src: Path, line_length: int, fast: bool, write_back: bool = False ) -> bool: """Format the file and rewrite if changed. Return True if changed.""" + 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 if write_back: - with open(src, "w", encoding=encoding) as f: + 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]: +def format_stdin_to_stdout( + line_length: int, fast: bool, write_back: bool = False +) -> bool: + """Format file on stdin and pipe output to stdout. Return True if changed.""" + 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.""" - with tokenize.open(src) as src_buffer: - src_contents = src_buffer.read() 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.""" 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): @@ -200,36 +258,41 @@ 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]) @@ -263,7 +326,7 @@ 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) @@ -285,6 +348,15 @@ 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} @@ -318,6 +390,7 @@ MATH_OPERATORS = { token.AMPER, token.PERCENT, token.CIRCUMFLEX, + token.TILDE, token.LEFTSHIFT, token.RIGHTSHIFT, token.DOUBLESTAR, @@ -333,10 +406,10 @@ 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) + 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: if leaf.type == token.COMMENT: @@ -345,8 +418,8 @@ class BracketTracker: 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: @@ -355,19 +428,25 @@ 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 @@ -377,7 +456,7 @@ 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_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. @@ -387,11 +466,13 @@ 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) + 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: has_value = leaf.value.strip() @@ -403,8 +484,10 @@ 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 @@ -426,9 +509,9 @@ class Line: @property def is_class(self) -> bool: 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 @@ -444,50 +527,55 @@ 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: 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: 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: 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 @@ -499,15 +587,42 @@ 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. @@ -518,8 +633,8 @@ 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 @@ -571,16 +686,55 @@ class Line: return bool(self.leaves or self.comments) +class UnformattedLines(Line): + + def append(self, leaf: Leaf, preformatted: bool = False) -> None: + 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: + raise NotImplementedError("Unformatted lines don't store comments separately.") + + def maybe_remove_trailing_comma(self, closing: Leaf) -> bool: + return False + + def maybe_increment_for_loop_variable(self, leaf: Leaf) -> bool: + return False + + def maybe_adapt_standalone_comment(self, comment: Leaf) -> bool: + return False + + def __str__(self) -> str: + 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`. @@ -589,17 +743,31 @@ 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.is_comment and 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(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: @@ -615,24 +783,23 @@ 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 @@ -646,10 +813,9 @@ 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. @@ -658,52 +824,68 @@ 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]: + """High-level entry point to the visitor.""" + 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]: 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]: 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]: + # 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]: @@ -737,7 +919,7 @@ class LineGenerator(Visitor[Line]): 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) @@ -756,6 +938,19 @@ class LineGenerator(Visitor[Line]): yield from self.visit_default(leaf) yield from self.line() + def visit_unformatted(self, node: LN) -> Iterator[Line]: + 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 @@ -775,9 +970,10 @@ 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 = ' ' @@ -785,56 +981,69 @@ def whitespace(leaf: Leaf) -> str: t = leaf.type p = leaf.parent v = leaf.value - if t == token.COLON: - return NO - - if t == token.COMMA: - return NO - - if t == token.RPAR: + if t in ALWAYS_NO_SPACE: return NO if t == token.COMMENT: return DOUBLESPACE - if t == STANDALONE_COMMENT: - return NO - - if t in CLOSING_BRACKETS: + 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 - assert p is not None, f"INTERNAL ERROR: hand-made leaf without parent: {leaf!r}" 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 and prevp.parent.type in { - syms.typedargslist, - syms.varargslist, - syms.parameters, - syms.arglist, - syms.argument, - }: - return NO + 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.typedargslist, - syms.varargslist, - syms.parameters, 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 == syms.subscript: + if prevp.parent and prevp.parent.type in {syms.subscript, syms.sliceop}: return NO - elif prevp.parent and prevp.parent.type == syms.factor: + elif ( + prevp.parent + and prevp.parent.type in {syms.factor, syms.star_expr} + and prevp.type in MATH_OPERATORS + ): + return NO + + 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: @@ -848,7 +1057,7 @@ def whitespace(leaf: Leaf) -> str: 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 @@ -931,7 +1140,14 @@ def whitespace(leaf: Leaf) -> str: elif p.type == syms.subscript: # indexing - if not prev or prev.type == token.COLON: + if not prev: + assert p.parent is not None, "subscripts are always parented" + if p.parent.type == syms.subscriptlist: + return SPACE + + return NO + + else: return NO elif p.type == syms.atom: @@ -940,9 +1156,9 @@ def whitespace(leaf: Leaf) -> str: 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 if not prev: @@ -956,7 +1172,7 @@ def whitespace(leaf: Leaf) -> str: 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 if not prev: prevp = preceding_leaf(p) @@ -1020,16 +1236,13 @@ 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 @@ -1055,40 +1268,68 @@ 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 split_line(line: Line, line_length: int, inner: bool = False) -> Iterator[Line]: +def make_comment(content: str) -> str: + 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 @@ -1110,11 +1351,13 @@ 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 @@ -1126,7 +1369,7 @@ def split_line(line: Line, line_length: int, inner: bool = False) -> Iterator[Li yield line -def left_hand_split(line: Line) -> Iterator[Line]: +def left_hand_split(line: Line, py36: bool = False) -> Iterator[Line]: """Split line into many lines, starting with the first matching bracket pair. Note: this usually looks weird, only use this for function definitions. @@ -1142,11 +1385,11 @@ 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: @@ -1154,7 +1397,7 @@ 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) @@ -1164,24 +1407,13 @@ 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" - ) - + 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]: +def right_hand_split(line: Line, py36: bool = False) -> Iterator[Line]: """Split 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) @@ -1194,18 +1426,18 @@ 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) @@ -1215,8 +1447,14 @@ 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')) + split_succeeded_or_raise(head, body, tail) + for result in (head, body, tail): + if result: + yield result + + +def split_succeeded_or_raise(head: Line, body: Line, tail: Line) -> None: + tail_len = len(str(tail).strip()) if not body: if tail_len == 0: raise CannotSplit("Splitting brackets produced the same line") @@ -1227,15 +1465,13 @@ 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]: +def delimiter_split(line: Line, py36: bool = False) -> Iterator[Line]: """Split 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] @@ -1249,24 +1485,34 @@ def delimiter_split(line: Line) -> Iterator[Line]: 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 @@ -1276,28 +1522,57 @@ 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: + """Leave existing extra newlines if not `inside_brackets`. + + Remove 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' @@ -1319,14 +1594,16 @@ 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) + 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) @@ -1345,7 +1622,7 @@ class Report: if self.failure_count: return 123 - elif self.change_count: + elif self.change_count and self.check: return 1 return 0 @@ -1355,21 +1632,27 @@ class Report: 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) + '.' @@ -1411,7 +1694,12 @@ 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) @@ -1420,7 +1708,7 @@ 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)) @@ -1431,7 +1719,7 @@ 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 @@ -1450,7 +1738,7 @@ 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