X-Git-Url: https://git.madduck.net/etc/vim.git/blobdiff_plain/1d3d743dfb71c7daaa9faa9186e00b053cd3cccb..c012c70176e1958d755768893c08f4b0892fb51d:/black.py diff --git a/black.py b/black.py index 01390b3..c77166a 100644 --- a/black.py +++ b/black.py @@ -1,6 +1,7 @@ #!/usr/bin/env python3 import asyncio +import pickle from asyncio.base_events import BaseEventLoop from concurrent.futures import Executor, ProcessPoolExecutor from enum import Enum @@ -10,18 +11,21 @@ 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, @@ -29,6 +33,7 @@ from typing import ( Union, ) +from appdirs import user_cache_dir from attr import dataclass, Factory import click @@ -38,7 +43,7 @@ from blib2to3 import pygram, pytree from blib2to3.pgen2 import driver, token from blib2to3.pgen2.parse import ParseError -__version__ = "18.4a0" +__version__ = "18.4a2" DEFAULT_LINE_LENGTH = 88 # types syms = pygram.python_symbols @@ -51,6 +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) @@ -101,6 +110,12 @@ class WriteBack(Enum): DIFF = 2 +class Changed(Enum): + NO = 0 + CACHED = 1 + YES = 2 + + @click.command() @click.option( "-l", @@ -169,12 +184,8 @@ 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: + if check and not diff: write_back = WriteBack.NO elif diff: write_back = WriteBack.DIFF @@ -182,22 +193,12 @@ def main( write_back = WriteBack.YES if len(sources) == 0: ctx.exit(0) + return + elif len(sources) == 1: - p = sources[0] - report = Report(check=check, quiet=quiet) - try: - if not p.is_file() and str(p) == "-": - changed = format_stdin_to_stdout( - line_length=line_length, fast=fast, write_back=write_back - ) - else: - changed = format_file_in_place( - p, line_length=line_length, fast=fast, write_back=write_back - ) - 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, check + ) else: loop = asyncio.get_event_loop() executor = ProcessPoolExecutor(max_workers=os.cpu_count()) @@ -205,12 +206,55 @@ def main( try: return_code = loop.run_until_complete( schedule_formatting( - sources, line_length, write_back, fast, quiet, loop, executor + sources, line_length, write_back, fast, quiet, loop, executor, check ) ) 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, + check: bool, +) -> 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=check, 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( @@ -221,6 +265,7 @@ async def schedule_formatting( quiet: bool, loop: BaseEventLoop, executor: Executor, + check: bool, ) -> int: """Run formatting of `sources` in parallel using the provided `executor`. @@ -229,41 +274,59 @@ async def schedule_formatting( `line_length`, `write_back`, and `fast` options are passed to :func:`format_file_in_place`. """ - 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()) - 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=check, 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=write_back is WriteBack.NO, quiet=quiet) - 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) elif not quiet: out("All done! ✨ 🍰 ✨") if not quiet: click.echo(str(report)) + + if write_back != WriteBack.DIFF and formatted: + write_cache(cache, formatted) + return report.return_code @@ -279,6 +342,7 @@ def format_file_in_place( 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: @@ -314,12 +378,12 @@ def format_stdin_to_stdout( `line_length` and `fast` arguments are passed to :func:`format_file_contents`. """ src = sys.stdin.read() + dst = src try: dst = format_file_contents(src, line_length=line_length, fast=fast) return True except NothingChanged: - dst = src return False finally: @@ -519,7 +583,20 @@ MATH_OPERATORS = { token.DOUBLESTAR, token.DOUBLESLASH, } -VARARGS = {token.STAR, token.DOUBLESTAR} +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 @@ -536,6 +613,8 @@ class BracketTracker: bracket_match: Dict[Tuple[Depth, NodeType], Leaf] = Factory(dict) delimiters: Dict[LeafID, Priority] = Factory(dict) previous: Optional[Leaf] = None + _for_loop_variable: bool = False + _lambda_arguments: bool = False def mark(self, leaf: Leaf) -> None: """Mark `leaf` with bracket-related metadata. Keep track of delimiters. @@ -555,22 +634,27 @@ class BracketTracker: if leaf.type == token.COMMENT: return + self.maybe_decrement_after_for_loop_variable(leaf) + self.maybe_decrement_after_lambda_arguments(leaf) if leaf.type in CLOSING_BRACKETS: self.depth -= 1 opening_bracket = self.bracket_match.pop((self.depth, leaf.type)) leaf.opening_bracket = opening_bracket leaf.bracket_depth = self.depth if self.depth == 0: - after_delim = is_split_after_delimiter(leaf, self.previous) - before_delim = is_split_before_delimiter(leaf, self.previous) - if after_delim > before_delim: - self.delimiters[id(leaf)] = after_delim - elif before_delim > after_delim and self.previous is not None: - self.delimiters[id(self.previous)] = before_delim + 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 self.previous = leaf + self.maybe_increment_lambda_arguments(leaf) + self.maybe_increment_for_loop_variable(leaf) def any_open_brackets(self) -> bool: """Return True if there is an yet unmatched open bracket on the line.""" @@ -579,10 +663,55 @@ 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) + 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, increase the depth of + tokens between `for` and `in`. + """ + if leaf.type == token.NAME and leaf.value == "for": + self.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.depth -= 1 + self._for_loop_variable = False + return True + + return False + + def maybe_increment_lambda_arguments(self, leaf: Leaf) -> bool: + """In a lambda expression, there might be more than one argument. + + To avoid splitting on the comma in this situation, increase the depth of + tokens between `lambda` and `:`. + """ + if leaf.type == token.NAME and leaf.value == "lambda": + self.depth += 1 + self._lambda_arguments = True + return True + + return False + + def maybe_decrement_after_lambda_arguments(self, leaf: Leaf) -> bool: + """See `maybe_increment_lambda_arguments` above for explanation.""" + if self._lambda_arguments and leaf.type == token.COLON: + self.depth -= 1 + self._lambda_arguments = False + return True + + return False + @dataclass class Line: @@ -593,8 +722,6 @@ class Line: comments: List[Tuple[Index, Leaf]] = Factory(list) 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. @@ -606,7 +733,7 @@ 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 @@ -615,10 +742,8 @@ 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 not self.append_comment(leaf): self.leaves.append(leaf) @@ -707,12 +832,12 @@ 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 @@ -765,29 +890,6 @@ class Line: 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, 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 append_comment(self, comment: Leaf) -> bool: """Add an inline or standalone comment to the line.""" if ( @@ -1075,15 +1177,22 @@ 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() @@ -1123,6 +1232,32 @@ 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() @@ -1153,18 +1288,23 @@ class LineGenerator(Visitor[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()) @@ -1212,15 +1352,8 @@ 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: @@ -1229,7 +1362,7 @@ 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 @@ -1250,17 +1383,11 @@ 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 @@ -1315,7 +1442,7 @@ 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: @@ -1354,21 +1481,9 @@ 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}: @@ -1380,9 +1495,10 @@ 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: @@ -1438,13 +1554,6 @@ def is_split_after_delimiter(leaf: Leaf, previous: Leaf = None) -> int: if leaf.type == token.COMMA: return COMMA_PRIORITY - if ( - leaf.type in VARARGS - and leaf.parent - and leaf.parent.type in {syms.argument, syms.typedargslist} - ): - return MATH_PRIORITY - return 0 @@ -1456,6 +1565,11 @@ def is_split_before_delimiter(leaf: Leaf, previous: Leaf = None) -> int: 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 and leaf.parent @@ -1495,17 +1609,6 @@ def is_split_before_delimiter(leaf: Leaf, previous: Leaf = None) -> int: return 0 -def is_delimiter(leaf: Leaf, previous: Leaf = None) -> int: - """Return the priority of the `leaf` delimiter. Return 0 if not delimiter. - - Higher numbers are higher priority. - """ - return max( - is_split_before_delimiter(leaf, previous), - is_split_after_delimiter(leaf, previous), - ) - - def generate_comments(leaf: Leaf) -> Iterator[Leaf]: """Clean the prefix of the `leaf` and generate comments from it, if any. @@ -1603,7 +1706,7 @@ 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 @@ -1669,9 +1772,7 @@ 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): @@ -1682,7 +1783,9 @@ 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) @@ -1692,14 +1795,16 @@ 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() @@ -1707,15 +1812,36 @@ 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 @@ -1806,8 +1932,7 @@ 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)) @@ -1829,12 +1954,7 @@ 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) @@ -1917,10 +2037,28 @@ 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] + '\\"' @@ -1932,10 +2070,142 @@ 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. @@ -1955,7 +2225,7 @@ 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 @@ -1991,16 +2261,20 @@ class Report: 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" if not self.quiet: out(f"{reformatted} {src}") self.change_count += 1 else: if not self.quiet: - out(f"{src} already well formatted, good job.", bold=False) + 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: @@ -2141,7 +2415,7 @@ 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) @@ -2190,5 +2464,71 @@ 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()