X-Git-Url: https://git.madduck.net/etc/vim.git/blobdiff_plain/80bd2b3134b4f01da4e279d040a224326b3577e5..b5b658da0683c3e0806461946d8b492784e26d97:/black.py diff --git a/black.py b/black.py index dc03e0a..9dd3536 100644 --- a/black.py +++ b/black.py @@ -1,24 +1,31 @@ #!/usr/bin/env python3 import asyncio +import pickle from asyncio.base_events import BaseEventLoop from concurrent.futures import Executor, ProcessPoolExecutor +from enum import Enum from functools import partial, wraps import keyword import logging +from multiprocessing import Manager import os from pathlib import Path +import re import tokenize import signal import sys from typing import ( + Any, Callable, + Collection, Dict, Generic, Iterable, Iterator, List, Optional, + Pattern, Set, Tuple, Type, @@ -26,6 +33,7 @@ from typing import ( Union, ) +from appdirs import user_cache_dir from attr import dataclass, Factory import click @@ -35,7 +43,7 @@ from blib2to3 import pygram, pytree from blib2to3.pgen2 import driver, token from blib2to3.pgen2.parse import ParseError -__version__ = "18.3a4" +__version__ = "18.4a2" DEFAULT_LINE_LENGTH = 88 # types syms = pygram.python_symbols @@ -48,6 +56,10 @@ Priority = int Index = int LN = Union[Leaf, Node] SplitFunc = Callable[["Line", bool], Iterator["Line"]] +Timestamp = float +FileSize = int +CacheInfo = Tuple[Timestamp, FileSize] +Cache = Dict[Path, CacheInfo] out = partial(click.secho, bold=True, err=True) err = partial(click.secho, fg="red", err=True) @@ -92,6 +104,18 @@ class FormatOff(FormatError): """Found a comment like `# fmt: off` in the file.""" +class WriteBack(Enum): + NO = 0 + YES = 1 + DIFF = 2 + + +class Changed(Enum): + NO = 0 + CACHED = 1 + YES = 2 + + @click.command() @click.option( "-l", @@ -105,16 +129,30 @@ class FormatOff(FormatError): "--check", is_flag=True, help=( - "Don't write back the files, just return the status. Return code 0 " + "Don't write the files back, just return the status. Return code 0 " "means nothing would change. Return code 1 means some files would be " "reformatted. Return code 123 means there was an internal error." ), ) +@click.option( + "--diff", + is_flag=True, + help="Don't write the files back, just output a diff for each file on stdout.", +) @click.option( "--fast/--safe", is_flag=True, help="If --fast given, skip temporary sanity checks. [default: --safe]", ) +@click.option( + "-q", + "--quiet", + is_flag=True, + help=( + "Don't emit non-error messages to stderr. Errors are still emitted, " + "silence those with 2>/dev/null." + ), +) @click.version_option(version=__version__) @click.argument( "src", @@ -125,7 +163,13 @@ class FormatOff(FormatError): ) @click.pass_context def main( - ctx: click.Context, line_length: int, check: bool, fast: bool, src: List[str] + ctx: click.Context, + line_length: int, + check: bool, + diff: bool, + fast: bool, + quiet: bool, + src: List[str], ) -> None: """The uncompromising code formatter.""" sources: List[Path] = [] @@ -140,24 +184,23 @@ def main( sources.append(Path("-")) else: err(f"invalid path: {s}") + if check and diff: + exc = click.ClickException("Options --check and --diff are mutually exclusive") + exc.exit_code = 2 + raise exc + + if check: + write_back = WriteBack.NO + elif diff: + write_back = WriteBack.DIFF + else: + write_back = WriteBack.YES if len(sources) == 0: ctx.exit(0) + return + elif len(sources) == 1: - p = sources[0] - report = Report(check=check) - try: - 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)) - ctx.exit(report.return_code) + return_code = reformat_one(sources[0], line_length, fast, quiet, write_back) else: loop = asyncio.get_event_loop() executor = ProcessPoolExecutor(max_workers=os.cpu_count()) @@ -165,19 +208,58 @@ def main( try: return_code = loop.run_until_complete( schedule_formatting( - sources, line_length, not check, fast, loop, executor + sources, line_length, write_back, fast, quiet, loop, executor ) ) finally: shutdown(loop) - ctx.exit(return_code) + ctx.exit(return_code) + + +def reformat_one( + src: Path, line_length: int, fast: bool, quiet: bool, write_back: WriteBack +) -> int: + """Reformat a single file under `src` without spawning child processes. + + If `quiet` is True, non-error messages are not output. `line_length`, + `write_back`, and `fast` options are passed to :func:`format_file_in_place`. + """ + report = Report(check=write_back is WriteBack.NO, quiet=quiet) + try: + changed = Changed.NO + if not src.is_file() and str(src) == "-": + if format_stdin_to_stdout( + line_length=line_length, fast=fast, write_back=write_back + ): + changed = Changed.YES + else: + cache: Cache = {} + if write_back != WriteBack.DIFF: + cache = read_cache() + src = src.resolve() + if src in cache and cache[src] == get_cache_info(src): + changed = Changed.CACHED + if ( + changed is not Changed.CACHED + and format_file_in_place( + src, line_length=line_length, fast=fast, write_back=write_back + ) + ): + changed = Changed.YES + if write_back != WriteBack.DIFF and changed is not Changed.NO: + write_cache(cache, [src]) + report.done(src, changed) + except Exception as exc: + report.failed(src, str(exc)) + return report.return_code async def schedule_formatting( sources: List[Path], line_length: int, - write_back: bool, + write_back: WriteBack, fast: bool, + quiet: bool, loop: BaseEventLoop, executor: Executor, ) -> int: @@ -188,79 +270,125 @@ async def schedule_formatting( `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, write_back - ) - for src in sources - } - _task_values = list(tasks.values()) - loop.add_signal_handler(signal.SIGINT, cancel, _task_values) - loop.add_signal_handler(signal.SIGTERM, cancel, _task_values) - await asyncio.wait(tasks.values()) + report = Report(check=write_back is WriteBack.NO, quiet=quiet) + cache: Cache = {} + if write_back != WriteBack.DIFF: + cache = read_cache() + sources, cached = filter_cached(cache, sources) + for src in cached: + report.done(src, Changed.CACHED) cancelled = [] - report = Report(check=not write_back) - for src, task in tasks.items(): - if not task.done(): - report.failed(src, "timed out, cancelling") - task.cancel() - cancelled.append(task) - elif task.cancelled(): - cancelled.append(task) - elif task.exception(): - report.failed(src, str(task.exception())) - else: - report.done(src, task.result()) + formatted = [] + if sources: + lock = None + if write_back == WriteBack.DIFF: + # For diff output, we need locks to ensure we don't interleave output + # from different processes. + manager = Manager() + lock = manager.Lock() + tasks = { + src: loop.run_in_executor( + executor, format_file_in_place, src, line_length, fast, write_back, lock + ) + for src in sources + } + _task_values = list(tasks.values()) + try: + loop.add_signal_handler(signal.SIGINT, cancel, _task_values) + loop.add_signal_handler(signal.SIGTERM, cancel, _task_values) + except NotImplementedError: + # There are no good alternatives for these on Windows + pass + await asyncio.wait(_task_values) + for src, task in tasks.items(): + if not task.done(): + report.failed(src, "timed out, cancelling") + task.cancel() + cancelled.append(task) + elif task.cancelled(): + cancelled.append(task) + elif task.exception(): + report.failed(src, str(task.exception())) + else: + formatted.append(src) + report.done(src, Changed.YES if task.result() else Changed.NO) + if cancelled: await asyncio.gather(*cancelled, loop=loop, return_exceptions=True) - else: + elif not quiet: out("All done! ✨ 🍰 ✨") - click.echo(str(report)) + if not quiet: + click.echo(str(report)) + + if write_back != WriteBack.DIFF and formatted: + write_cache(cache, formatted) + return report.return_code def format_file_in_place( - src: Path, line_length: int, fast: bool, write_back: bool = False + src: Path, + line_length: int, + fast: bool, + write_back: WriteBack = WriteBack.NO, + lock: Any = None, # multiprocessing.Manager().Lock() is some crazy proxy ) -> 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 = format_file_contents( + dst_contents = format_file_contents( src_contents, line_length=line_length, fast=fast ) except NothingChanged: return False - if write_back: + if write_back == write_back.YES: with open(src, "w", encoding=src_buffer.encoding) as f: - f.write(contents) + f.write(dst_contents) + elif write_back == write_back.DIFF: + src_name = f"{src.name} (original)" + dst_name = f"{src.name} (formatted)" + diff_contents = diff(src_contents, dst_contents, src_name, dst_name) + if lock: + lock.acquire() + try: + sys.stdout.write(diff_contents) + finally: + if lock: + lock.release() return True def format_stdin_to_stdout( - line_length: int, fast: bool, write_back: bool = False + line_length: int, fast: bool, write_back: WriteBack = WriteBack.NO ) -> 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() + src = sys.stdin.read() + dst = src try: - contents = format_file_contents(contents, line_length=line_length, fast=fast) + dst = format_file_contents(src, line_length=line_length, fast=fast) return True except NothingChanged: return False finally: - if write_back: - sys.stdout.write(contents) + if write_back == WriteBack.YES: + sys.stdout.write(dst) + elif write_back == WriteBack.DIFF: + src_name = " (original)" + dst_name = " (formatted)" + sys.stdout.write(diff(src, dst, src_name, dst_name)) def format_file_contents( @@ -451,6 +579,20 @@ MATH_OPERATORS = { token.DOUBLESTAR, token.DOUBLESLASH, } +STARS = {token.STAR, token.DOUBLESTAR} +VARARGS_PARENTS = { + syms.arglist, + syms.argument, # double star in arglist + syms.trailer, # single argument to call + syms.typedargslist, + syms.varargslist, # lambdas +} +UNPACKING_PARENTS = { + syms.atom, # single element of a list or set literal + syms.dictsetmaker, + syms.listmaker, + syms.testlist_gexp, +} COMPREHENSION_PRIORITY = 20 COMMA_PRIORITY = 10 LOGIC_PRIORITY = 5 @@ -492,32 +634,13 @@ class BracketTracker: leaf.opening_bracket = opening_bracket leaf.bracket_depth = self.depth if self.depth == 0: - delim = is_delimiter(leaf) - if delim: - self.delimiters[id(leaf)] = delim - elif self.previous is not None: - 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} - ): - 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} - ): - 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 + delim = is_split_before_delimiter(leaf, self.previous) + if delim and self.previous is not None: + self.delimiters[id(self.previous)] = delim + else: + delim = is_split_after_delimiter(leaf, self.previous) + if delim: + self.delimiters[id(leaf)] = delim if leaf.type in OPENING_BRACKETS: self.bracket_match[self.depth, BRACKET[leaf.type]] = leaf self.depth += 1 @@ -530,7 +653,8 @@ class BracketTracker: def max_delimiter_priority(self, exclude: Iterable[LeafID] = ()) -> int: """Return the highest priority of a delimiter found on the line. - Values are consistent with what `is_delimiter()` returns. + Values are consistent with what `is_split_*_delimiter()` return. + Raises ValueError on no delimiters. """ return max(v for k, v in self.delimiters.items() if k not in exclude) @@ -557,7 +681,7 @@ class Line: Inline comments are put aside. """ - has_value = leaf.value.strip() + has_value = leaf.type in BRACKETS or bool(leaf.value.strip()) if not has_value: return @@ -658,12 +782,12 @@ class Line: and self.leaves[0].value == "yield" ) - @property - def contains_standalone_comments(self) -> bool: + def contains_standalone_comments(self, depth_limit: int = sys.maxsize) -> bool: """If so, needs to be split before emitting.""" for leaf in self.leaves: if leaf.type == STANDALONE_COMMENT: - return True + if leaf.bracket_depth <= depth_limit: + return True return False @@ -1026,15 +1150,22 @@ class LineGenerator(Visitor[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]: + def visit_stmt( + self, node: Node, keywords: Set[str], parens: Set[str] + ) -> Iterator[Line]: """Visit a statement. This implementation is shared for `if`, `while`, `for`, `try`, `except`, - `def`, `with`, and `class`. + `def`, `with`, `class`, and `assert`. + + The relevant Python language `keywords` for a given statement will be + NAME leaves within it. This methods puts those on a separate line. - The relevant Python language `keywords` for a given statement will be NAME - leaves within it. This methods puts those on a separate line. + `parens` holds pairs of nodes where invisible parentheses should be put. + Keys hold nodes after which opening parentheses should be put, values + hold nodes before which closing parentheses should be put. """ + normalize_invisible_parens(node, parens_after=parens) for child in node.children: if child.type == token.NAME and child.value in keywords: # type: ignore yield from self.line() @@ -1074,6 +1205,32 @@ class LineGenerator(Visitor[Line]): yield from self.line() yield from self.visit(child) + def visit_import_from(self, node: Node) -> Iterator[Line]: + """Visit import_from and maybe put invisible parentheses. + + This is separate from `visit_stmt` because import statements don't + support arbitrary atoms and thus handling of parentheses is custom. + """ + check_lpar = False + for index, child in enumerate(node.children): + if check_lpar: + if child.type == token.LPAR: + # make parentheses invisible + child.value = "" # type: ignore + node.children[-1].value = "" # type: ignore + else: + # insert invisible parentheses + node.insert_child(index, Leaf(token.LPAR, "")) + node.append_child(Leaf(token.RPAR, "")) + break + + check_lpar = ( + child.type == token.NAME and child.value == "import" # type: ignore + ) + + for child in node.children: + yield from self.visit(child) + def visit_SEMI(self, leaf: Leaf) -> Iterator[Line]: """Remove a semicolon and put the other statement on a separate line.""" yield from self.line() @@ -1097,21 +1254,30 @@ class LineGenerator(Visitor[Line]): yield from self.line() yield from self.visit(node) + if node.type == token.ENDMARKER: + # somebody decided not to put a final `# fmt: on` + yield from self.line() + def __attrs_post_init__(self) -> None: """You are in a twisty little maze of passages.""" v = self.visit_stmt - self.visit_if_stmt = partial(v, keywords={"if", "else", "elif"}) - self.visit_while_stmt = partial(v, keywords={"while", "else"}) - self.visit_for_stmt = partial(v, keywords={"for", "else"}) - self.visit_try_stmt = partial(v, keywords={"try", "except", "else", "finally"}) - self.visit_except_clause = partial(v, keywords={"except"}) - self.visit_funcdef = partial(v, keywords={"def"}) - self.visit_with_stmt = partial(v, keywords={"with"}) - self.visit_classdef = partial(v, keywords={"class"}) + Ø: Set[str] = set() + self.visit_assert_stmt = partial(v, keywords={"assert"}, parens={"assert", ","}) + self.visit_if_stmt = partial(v, keywords={"if", "else", "elif"}, parens={"if"}) + self.visit_while_stmt = partial(v, keywords={"while", "else"}, parens={"while"}) + self.visit_for_stmt = partial(v, keywords={"for", "else"}, parens={"for", "in"}) + self.visit_try_stmt = partial( + v, keywords={"try", "except", "else", "finally"}, parens=Ø + ) + self.visit_except_clause = partial(v, keywords={"except"}, parens=Ø) + self.visit_with_stmt = partial(v, keywords={"with"}, parens=Ø) + self.visit_funcdef = partial(v, keywords={"def"}, parens=Ø) + self.visit_classdef = partial(v, keywords={"class"}, parens=Ø) self.visit_async_funcdef = self.visit_async_stmt self.visit_decorated = self.visit_decorators +IMPLICIT_TUPLE = {syms.testlist, syms.testlist_star_expr, syms.exprlist} BRACKET = {token.LPAR: token.RPAR, token.LSQB: token.RSQB, token.LBRACE: token.RBRACE} OPENING_BRACKETS = set(BRACKET.keys()) CLOSING_BRACKETS = set(BRACKET.values()) @@ -1159,15 +1325,8 @@ def whitespace(leaf: Leaf) -> str: # noqa C901 # 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, - }: + elif prevp.type in STARS: + if is_vararg(prevp, within=VARARGS_PARENTS | UNPACKING_PARENTS): return NO elif prevp.type == token.COLON: @@ -1176,7 +1335,7 @@ def whitespace(leaf: Leaf) -> str: # noqa C901 elif ( prevp.parent - and prevp.parent.type in {syms.factor, syms.star_expr} + and prevp.parent.type == syms.factor and prevp.type in MATH_OPERATORS ): return NO @@ -1197,17 +1356,11 @@ def whitespace(leaf: Leaf) -> str: # noqa C901 if p.type in {syms.parameters, syms.arglist}: # untyped function signatures or calls - if t == token.RPAR: - return NO - if not prev or prev.type != token.COMMA: return NO elif p.type == syms.varargslist: # lambdas - if t == token.RPAR: - return NO - if prev and prev.type != token.COMMA: return NO @@ -1262,7 +1415,7 @@ def whitespace(leaf: Leaf) -> str: # noqa C901 if not prevp or prevp.type == token.LPAR: return NO - elif prev.type == token.EQUAL or prev.type == token.DOUBLESTAR: + elif prev.type in {token.EQUAL} | STARS: return NO elif p.type == syms.decorator: @@ -1301,21 +1454,9 @@ def whitespace(leaf: Leaf) -> str: # noqa C901 # dots, but not the first one. return NO - elif ( - p.type == syms.listmaker - or p.type == syms.testlist_gexp - or p.type == syms.subscriptlist - ): - # list interior, including unpacking - if not prev: - return NO - elif p.type == syms.dictsetmaker: - # dict and set interior, including unpacking - if not prev: - return NO - - if prev.type == token.DOUBLESTAR: + # dict unpacking + if prev and prev.type == token.DOUBLESTAR: return NO elif p.type in {syms.factor, syms.star_expr}: @@ -1327,9 +1468,10 @@ def whitespace(leaf: Leaf) -> str: # noqa C901 prevp_parent = prevp.parent assert prevp_parent is not None - if prevp.type == token.COLON and prevp_parent.type in { - syms.subscript, syms.sliceop - }: + if ( + prevp.type == token.COLON + and prevp_parent.type in {syms.subscript, syms.sliceop} + ): return NO elif prevp.type == token.EQUAL and prevp_parent.type == syms.argument: @@ -1374,16 +1516,32 @@ def preceding_leaf(node: Optional[LN]) -> Optional[Leaf]: return None -def is_delimiter(leaf: Leaf) -> int: - """Return the priority of the `leaf` delimiter. Return 0 if not delimiter. +def is_split_after_delimiter(leaf: Leaf, previous: Leaf = None) -> int: + """Return the priority of the `leaf` delimiter, given a line break after it. + + The delimiter priorities returned here are from those delimiters that would + cause a line break after themselves. Higher numbers are higher priority. """ if leaf.type == token.COMMA: return COMMA_PRIORITY - if leaf.type in COMPARATORS: - return COMPARATOR_PRIORITY + return 0 + + +def is_split_before_delimiter(leaf: Leaf, previous: Leaf = None) -> int: + """Return the priority of the `leaf` delimiter, given a line before after it. + + The delimiter priorities returned here are from those delimiters that would + cause a line break before themselves. + + Higher numbers are higher priority. + """ + if is_vararg(leaf, within=VARARGS_PARENTS | UNPACKING_PARENTS): + # * and ** might also be MATH_OPERATORS but in this case they are not. + # Don't treat them as a delimiter. + return 0 if ( leaf.type in MATH_OPERATORS @@ -1392,6 +1550,35 @@ def is_delimiter(leaf: Leaf) -> int: ): return MATH_PRIORITY + if leaf.type in COMPARATORS: + return COMPARATOR_PRIORITY + + if ( + leaf.type == token.STRING + and previous is not None + and previous.type == token.STRING + ): + return STRING_PRIORITY + + if ( + leaf.type == token.NAME + and leaf.value == "for" + and leaf.parent + and leaf.parent.type in {syms.comp_for, syms.old_comp_for} + ): + return COMPREHENSION_PRIORITY + + if ( + leaf.type == token.NAME + and leaf.value == "if" + and leaf.parent + and leaf.parent.type in {syms.comp_if, syms.old_comp_if} + ): + return COMPREHENSION_PRIORITY + + if leaf.type == token.NAME and leaf.value in LOGIC_OPERATORS and leaf.parent: + return LOGIC_PRIORITY + return 0 @@ -1442,7 +1629,12 @@ def generate_comments(leaf: Leaf) -> Iterator[Leaf]: raise FormatOn(consumed) if comment in {"# fmt: off", "# yapf: disable"}: - raise FormatOff(consumed) + if comment_type == STANDALONE_COMMENT: + raise FormatOff(consumed) + + prev = preceding_leaf(leaf) + if not prev or prev.type in WHITESPACE: # standalone comment in disguise + raise FormatOff(consumed) nlines = 0 @@ -1487,7 +1679,7 @@ def split_line( if ( len(line_str) <= line_length and "\n" not in line_str # multiline strings - and not line.contains_standalone_comments + and not line.contains_standalone_comments() ): yield line return @@ -1553,9 +1745,7 @@ def left_hand_split(line: Line, py36: bool = False) -> Iterator[Line]: if body_leaves: 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) - ): + for result, leaves in (head, head_leaves), (body, body_leaves), (tail, tail_leaves): for leaf in leaves: result.append(leaf, preformatted=True) for comment_after in line.comments_after(leaf): @@ -1566,7 +1756,9 @@ def left_hand_split(line: Line, py36: bool = False) -> Iterator[Line]: yield result -def right_hand_split(line: Line, py36: bool = False) -> Iterator[Line]: +def right_hand_split( + line: Line, py36: bool = False, omit: Collection[LeafID] = () +) -> 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) @@ -1576,14 +1768,16 @@ def right_hand_split(line: Line, py36: bool = False) -> Iterator[Line]: head_leaves: List[Leaf] = [] current_leaves = tail_leaves opening_bracket = None + closing_bracket = None for leaf in reversed(line.leaves): if current_leaves is body_leaves: if leaf is opening_bracket: 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: + if leaf.type in CLOSING_BRACKETS and id(leaf) not in omit: opening_bracket = leaf.opening_bracket + closing_bracket = leaf current_leaves = body_leaves tail_leaves.reverse() body_leaves.reverse() @@ -1591,15 +1785,36 @@ def right_hand_split(line: Line, py36: bool = False) -> Iterator[Line]: # Since body is a new indent level, remove spurious leading whitespace. if body_leaves: normalize_prefix(body_leaves[0], inside_brackets=True) + elif not head_leaves: + # No `head` and no `body` means the split failed. `tail` has all content. + raise CannotSplit("No brackets found") + # Build the new lines. - for result, leaves in ( - (head, head_leaves), (body, body_leaves), (tail, tail_leaves) - ): + for result, leaves in (head, head_leaves), (body, body_leaves), (tail, tail_leaves): for leaf in leaves: result.append(leaf, preformatted=True) for comment_after in line.comments_after(leaf): result.append(comment_after, preformatted=True) bracket_split_succeeded_or_raise(head, body, tail) + assert opening_bracket and closing_bracket + if ( + opening_bracket.type == token.LPAR + and not opening_bracket.value + and closing_bracket.type == token.RPAR + and not closing_bracket.value + ): + # These parens were optional. If there aren't any delimiters or standalone + # comments in the body, they were unnecessary and another split without + # them should be attempted. + if not ( + body.bracket_tracker.delimiters or line.contains_standalone_comments(0) + ): + omit = {id(closing_bracket), *omit} + yield from right_hand_split(line, py36=py36, omit=omit) + return + + ensure_visible(opening_bracket) + ensure_visible(closing_bracket) for result in (head, body, tail): if result: yield result @@ -1690,8 +1905,7 @@ def delimiter_split(line: Line, py36: bool = False) -> Iterator[Line]: lowest_depth = min(lowest_depth, leaf.bracket_depth) if ( leaf.bracket_depth == lowest_depth - and leaf.type == token.STAR - or leaf.type == token.DOUBLESTAR + and is_vararg(leaf, within=VARARGS_PARENTS) ): trailing_comma_safe = trailing_comma_safe and py36 leaf_priority = delimiters.get(id(leaf)) @@ -1701,9 +1915,10 @@ def delimiter_split(line: Line, py36: bool = False) -> Iterator[Line]: current_line = Line(depth=line.depth, inside_brackets=line.inside_brackets) if current_line: if ( - delimiter_priority == COMMA_PRIORITY + trailing_comma_safe + and delimiter_priority == COMMA_PRIORITY and current_line.leaves[-1].type != token.COMMA - and trailing_comma_safe + and current_line.leaves[-1].type != STANDALONE_COMMENT ): current_line.append(Leaf(token.COMMA, ",")) yield current_line @@ -1712,12 +1927,7 @@ def delimiter_split(line: Line, py36: bool = False) -> Iterator[Line]: @dont_increase_indentation def standalone_comment_split(line: Line, py36: bool = False) -> Iterator[Line]: """Split standalone comments from the rest of the line.""" - for leaf in line.leaves: - if leaf.type == STANDALONE_COMMENT: - if leaf.bracket_depth == 0: - break - - else: + if not line.contains_standalone_comments(0): raise CannotSplit("Line does not have any standalone comments") current_line = Line(depth=line.depth, inside_brackets=line.inside_brackets) @@ -1776,6 +1986,13 @@ def normalize_prefix(leaf: Leaf, *, inside_brackets: bool) -> None: def normalize_string_quotes(leaf: Leaf) -> None: + """Prefer double quotes but only if it doesn't cause more escaping. + + Adds or removes backslashes as appropriate. Doesn't parse and fix + strings nested in f-strings (yet). + + Note: Mutates its argument. + """ value = leaf.value.lstrip("furbFURB") if value[:3] == '"""': return @@ -1793,10 +2010,28 @@ def normalize_string_quotes(leaf: Leaf) -> None: if first_quote_pos == -1: return # There's an internal error + prefix = leaf.value[:first_quote_pos] + unescaped_new_quote = re.compile(rf"(([^\\]|^)(\\\\)*){new_quote}") + escaped_new_quote = re.compile(rf"([^\\]|^)\\(\\\\)*{new_quote}") + escaped_orig_quote = re.compile(rf"([^\\]|^)\\(\\\\)*{orig_quote}") body = leaf.value[first_quote_pos + len(orig_quote):-len(orig_quote)] - new_body = body.replace(f"\\{orig_quote}", orig_quote).replace( - new_quote, f"\\{new_quote}" - ) + if "r" in prefix.casefold(): + if unescaped_new_quote.search(body): + # There's at least one unescaped new_quote in this raw string + # so converting is impossible + return + + # Do not introduce or remove backslashes in raw strings + new_body = body + else: + # remove unnecessary quotes + new_body = sub_twice(escaped_new_quote, rf"\1\2{new_quote}", body) + if body != new_body: + # Consider the string without unnecessary quotes as the original + body = new_body + leaf.value = f"{prefix}{orig_quote}{body}{orig_quote}" + new_body = sub_twice(escaped_orig_quote, rf"\1\2{orig_quote}", new_body) + new_body = sub_twice(unescaped_new_quote, rf"\1\\{new_quote}", new_body) if new_quote == '"""' and new_body[-1] == '"': # edge case: new_body = new_body[:-1] + '\\"' @@ -1808,10 +2043,142 @@ def normalize_string_quotes(leaf: Leaf) -> None: if new_escape_count == orig_escape_count and orig_quote == '"': return # Prefer double quotes - prefix = leaf.value[:first_quote_pos] leaf.value = f"{prefix}{new_quote}{new_body}{new_quote}" +def normalize_invisible_parens(node: Node, parens_after: Set[str]) -> None: + """Make existing optional parentheses invisible or create new ones. + + Standardizes on visible parentheses for single-element tuples, and keeps + existing visible parentheses for other tuples and generator expressions. + """ + check_lpar = False + for child in list(node.children): + if check_lpar: + if child.type == syms.atom: + if not ( + is_empty_tuple(child) + or is_one_tuple(child) + or max_delimiter_priority_in_atom(child) >= COMMA_PRIORITY + ): + first = child.children[0] + last = child.children[-1] + if first.type == token.LPAR and last.type == token.RPAR: + # make parentheses invisible + first.value = "" # type: ignore + last.value = "" # type: ignore + elif is_one_tuple(child): + # wrap child in visible parentheses + lpar = Leaf(token.LPAR, "(") + rpar = Leaf(token.RPAR, ")") + index = child.remove() or 0 + node.insert_child(index, Node(syms.atom, [lpar, child, rpar])) + else: + # wrap child in invisible parentheses + lpar = Leaf(token.LPAR, "") + rpar = Leaf(token.RPAR, "") + index = child.remove() or 0 + node.insert_child(index, Node(syms.atom, [lpar, child, rpar])) + + check_lpar = isinstance(child, Leaf) and child.value in parens_after + + +def is_empty_tuple(node: LN) -> bool: + """Return True if `node` holds an empty tuple.""" + return ( + node.type == syms.atom + and len(node.children) == 2 + and node.children[0].type == token.LPAR + and node.children[1].type == token.RPAR + ) + + +def is_one_tuple(node: LN) -> bool: + """Return True if `node` holds a tuple with one element, with or without parens.""" + if node.type == syms.atom: + if len(node.children) != 3: + return False + + lpar, gexp, rpar = node.children + if not ( + lpar.type == token.LPAR + and gexp.type == syms.testlist_gexp + and rpar.type == token.RPAR + ): + return False + + return len(gexp.children) == 2 and gexp.children[1].type == token.COMMA + + return ( + node.type in IMPLICIT_TUPLE + and len(node.children) == 2 + and node.children[1].type == token.COMMA + ) + + +def is_vararg(leaf: Leaf, within: Set[NodeType]) -> bool: + """Return True if `leaf` is a star or double star in a vararg or kwarg. + + If `within` includes VARARGS_PARENTS, this applies to function signatures. + If `within` includes COLLECTION_LIBERALS_PARENTS, it applies to right + hand-side extended iterable unpacking (PEP 3132) and additional unpacking + generalizations (PEP 448). + """ + if leaf.type not in STARS or not leaf.parent: + return False + + p = leaf.parent + if p.type == syms.star_expr: + # Star expressions are also used as assignment targets in extended + # iterable unpacking (PEP 3132). See what its parent is instead. + if not p.parent: + return False + + p = p.parent + + return p.type in within + + +def max_delimiter_priority_in_atom(node: LN) -> int: + """Return maximum delimiter priority inside `node`. + + This is specific to atoms with contents contained in a pair of parentheses. + If `node` isn't an atom or there are no enclosing parentheses, returns 0. + """ + if node.type != syms.atom: + return 0 + + first = node.children[0] + last = node.children[-1] + if not (first.type == token.LPAR and last.type == token.RPAR): + return 0 + + bt = BracketTracker() + for c in node.children[1:-1]: + if isinstance(c, Leaf): + bt.mark(c) + else: + for leaf in c.leaves(): + bt.mark(leaf) + try: + return bt.max_delimiter_priority() + + except ValueError: + return 0 + + +def ensure_visible(leaf: Leaf) -> None: + """Make sure parentheses are visible. + + They could be invisible as part of some statements (see + :func:`normalize_invible_parens` and :func:`visit_import_from`). + """ + if leaf.type == token.LPAR: + leaf.value = "(" + elif leaf.type == token.RPAR: + leaf.value = ")" + + def is_python36(node: Node) -> bool: """Return True if the current file is using Python 3.6+ features. @@ -1831,7 +2198,7 @@ def is_python36(node: Node) -> bool: and n.children[-1].type == token.COMMA ): for ch in n.children: - if ch.type == token.STAR or ch.type == token.DOUBLESTAR: + if ch.type in STARS: return True return False @@ -1862,18 +2229,25 @@ def gen_python_files_in_dir(path: Path) -> Iterator[Path]: class Report: """Provides a reformatting counter. Can be rendered with `str(report)`.""" check: bool = False + quiet: bool = False change_count: int = 0 same_count: int = 0 failure_count: int = 0 - def done(self, src: Path, changed: bool) -> None: + def done(self, src: Path, changed: Changed) -> None: """Increment the counter for successful reformatting. Write out a message.""" - if changed: + if changed is Changed.YES: reformatted = "would reformat" if self.check else "reformatted" - out(f"{reformatted} {src}") + if not self.quiet: + out(f"{reformatted} {src}") self.change_count += 1 else: - out(f"{src} already well formatted, good job.", bold=False) + if not self.quiet: + if changed is Changed.NO: + msg = f"{src} already well formatted, good job." + else: + msg = f"{src} wasn't modified on disk since last run." + out(msg, bold=False) self.same_count += 1 def failed(self, src: Path, message: str) -> None: @@ -2014,11 +2388,12 @@ def dump_to_file(*output: str) -> str: import tempfile with tempfile.NamedTemporaryFile( - mode="w", prefix="blk_", suffix=".log", delete=False + mode="w", prefix="blk_", suffix=".log", delete=False, encoding="utf8" ) as f: for lines in output: f.write(lines) - f.write("\n") + if lines and lines[-1] != "\n": + f.write("\n") return f.name @@ -2062,5 +2437,71 @@ def shutdown(loop: BaseEventLoop) -> None: loop.close() +def sub_twice(regex: Pattern[str], replacement: str, original: str) -> str: + """Replace `regex` with `replacement` twice on `original`. + + This is used by string normalization to perform replaces on + overlapping matches. + """ + return regex.sub(replacement, regex.sub(replacement, original)) + + +CACHE_DIR = Path(user_cache_dir("black", version=__version__)) +CACHE_FILE = CACHE_DIR / "cache.pickle" + + +def read_cache() -> Cache: + """Read the cache if it exists and is well formed. + + If it is not well formed, the call to write_cache later should resolve the issue. + """ + if not CACHE_FILE.exists(): + return {} + + with CACHE_FILE.open("rb") as fobj: + try: + cache: Cache = pickle.load(fobj) + except pickle.UnpicklingError: + return {} + + return cache + + +def get_cache_info(path: Path) -> CacheInfo: + """Return the information used to check if a file is already formatted or not.""" + stat = path.stat() + return stat.st_mtime, stat.st_size + + +def filter_cached( + cache: Cache, sources: Iterable[Path] +) -> Tuple[List[Path], List[Path]]: + """Split a list of paths into two. + + The first list contains paths of files that modified on disk or are not in the + cache. The other list contains paths to non-modified files. + """ + todo, done = [], [] + for src in sources: + src = src.resolve() + if cache.get(src) != get_cache_info(src): + todo.append(src) + else: + done.append(src) + return todo, done + + +def write_cache(cache: Cache, sources: List[Path]) -> None: + """Update the cache file.""" + try: + if not CACHE_DIR.exists(): + CACHE_DIR.mkdir(parents=True) + new_cache = {**cache, **{src.resolve(): get_cache_info(src) for src in sources}} + with CACHE_FILE.open("wb") as fobj: + pickle.dump(new_cache, fobj, protocol=pickle.HIGHEST_PROTOCOL) + except OSError: + pass + + if __name__ == "__main__": main()