X-Git-Url: https://git.madduck.net/etc/vim.git/blobdiff_plain/36bed855e11b119adc4cd5b3ad87e2da965928ba..66b82ced502a826649f9c7172a9a1d24e415d792:/black.py?ds=inline diff --git a/black.py b/black.py index 730c64d..c899bde 100644 --- a/black.py +++ b/black.py @@ -1,23 +1,26 @@ import asyncio -import pickle from asyncio.base_events import BaseEventLoop from concurrent.futures import Executor, ProcessPoolExecutor +from datetime import datetime from enum import Enum, Flag -from functools import partial, wraps +from functools import lru_cache, partial, wraps +import io import keyword import logging from multiprocessing import Manager import os from pathlib import Path +import pickle import re -import tokenize import signal import sys +import tokenize from typing import ( Any, Callable, Collection, Dict, + Generator, Generic, Iterable, Iterator, @@ -27,7 +30,6 @@ from typing import ( Sequence, Set, Tuple, - Type, TypeVar, Union, cast, @@ -36,6 +38,7 @@ from typing import ( from appdirs import user_cache_dir from attr import dataclass, Factory import click +import toml # lib2to3 fork from blib2to3.pytree import Node, Leaf, type_repr @@ -44,7 +47,7 @@ from blib2to3.pgen2 import driver, token from blib2to3.pgen2.parse import ParseError -__version__ = "18.5b1" +__version__ = "18.6b4" DEFAULT_LINE_LENGTH = 88 DEFAULT_EXCLUDES = ( r"/(\.git|\.hg|\.mypy_cache|\.tox|\.venv|_build|buck-out|build|dist)/" @@ -56,6 +59,7 @@ CACHE_DIR = Path(user_cache_dir("black", version=__version__)) # types FileContent = str Encoding = str +NewLine = str Depth = int NodeType = int LeafID = int @@ -86,38 +90,18 @@ class CannotSplit(Exception): """ -class FormatError(Exception): - """Base exception for `# fmt: on` and `# fmt: off` handling. - - It holds the number of bytes of the prefix consumed before the format - control comment appeared. - """ - - def __init__(self, consumed: int) -> None: - super().__init__(consumed) - self.consumed = consumed - - def trim_prefix(self, leaf: Leaf) -> None: - leaf.prefix = leaf.prefix[self.consumed :] - - def leaf_from_consumed(self, leaf: Leaf) -> Leaf: - """Returns a new Leaf from the consumed part of the prefix.""" - unformatted_prefix = leaf.prefix[: self.consumed] - return Leaf(token.NEWLINE, unformatted_prefix) - - -class FormatOn(FormatError): - """Found a comment like `# fmt: on` in the file.""" - - -class FormatOff(FormatError): - """Found a comment like `# fmt: off` in the file.""" - - class WriteBack(Enum): NO = 0 YES = 1 DIFF = 2 + CHECK = 3 + + @classmethod + def from_configuration(cls, *, check: bool, diff: bool) -> "WriteBack": + if check and not diff: + return cls.CHECK + + return cls.DIFF if diff else cls.YES class Changed(Enum): @@ -132,14 +116,61 @@ class FileMode(Flag): PYI = 2 NO_STRING_NORMALIZATION = 4 + @classmethod + def from_configuration( + cls, *, py36: bool, pyi: bool, skip_string_normalization: bool + ) -> "FileMode": + mode = cls.AUTO_DETECT + if py36: + mode |= cls.PYTHON36 + if pyi: + mode |= cls.PYI + if skip_string_normalization: + mode |= cls.NO_STRING_NORMALIZATION + return mode + + +def read_pyproject_toml( + ctx: click.Context, param: click.Parameter, value: Union[str, int, bool, None] +) -> Optional[str]: + """Inject Black configuration from "pyproject.toml" into defaults in `ctx`. + + Returns the path to a successfully found and read configuration file, None + otherwise. + """ + assert not isinstance(value, (int, bool)), "Invalid parameter type passed" + if not value: + root = find_project_root(ctx.params.get("src", ())) + path = root / "pyproject.toml" + if path.is_file(): + value = str(path) + else: + return None -@click.command() + try: + pyproject_toml = toml.load(value) + config = pyproject_toml.get("tool", {}).get("black", {}) + except (toml.TomlDecodeError, OSError) as e: + raise click.BadOptionUsage(f"Error reading configuration file: {e}", ctx) + + if not config: + return None + + if ctx.default_map is None: + ctx.default_map = {} + ctx.default_map.update( # type: ignore # bad types in .pyi + {k.replace("--", "").replace("-", "_"): v for k, v in config.items()} + ) + return value + + +@click.command(context_settings=dict(help_option_names=["-h", "--help"])) @click.option( "-l", "--line-length", type=int, default=DEFAULT_LINE_LENGTH, - help="How many character per line to allow.", + help="How many characters per line to allow.", show_default=True, ) @click.option( @@ -218,6 +249,15 @@ class FileMode(Flag): "silence those with 2>/dev/null." ), ) +@click.option( + "-v", + "--verbose", + is_flag=True, + help=( + "Also emit messages to stderr about files that were not changed or were " + "ignored due to --exclude=." + ), +) @click.version_option(version=__version__) @click.argument( "src", @@ -225,6 +265,16 @@ class FileMode(Flag): type=click.Path( exists=True, file_okay=True, dir_okay=True, readable=True, allow_dash=True ), + is_eager=True, +) +@click.option( + "--config", + type=click.Path( + exists=False, file_okay=True, dir_okay=False, readable=True, allow_dash=False + ), + is_eager=True, + callback=read_pyproject_toml, + help="Read configuration from PATH.", ) @click.pass_context def main( @@ -237,59 +287,51 @@ def main( py36: bool, skip_string_normalization: bool, quiet: bool, + verbose: bool, include: str, exclude: str, - src: List[str], + src: Tuple[str], + config: Optional[str], ) -> None: """The uncompromising code formatter.""" - sources: List[Path] = [] + write_back = WriteBack.from_configuration(check=check, diff=diff) + mode = FileMode.from_configuration( + py36=py36, pyi=pyi, skip_string_normalization=skip_string_normalization + ) + if config and verbose: + out(f"Using configuration from {config}.", bold=False, fg="blue") try: - include_regex = re.compile(include) + include_regex = re_compile_maybe_verbose(include) except re.error: err(f"Invalid regular expression for include given: {include!r}") ctx.exit(2) try: - exclude_regex = re.compile(exclude) + exclude_regex = re_compile_maybe_verbose(exclude) except re.error: err(f"Invalid regular expression for exclude given: {exclude!r}") ctx.exit(2) + report = Report(check=check, quiet=quiet, verbose=verbose) root = find_project_root(src) + sources: Set[Path] = set() for s in src: p = Path(s) if p.is_dir(): - sources.extend( - gen_python_files_in_dir(p, root, include_regex, exclude_regex) + sources.update( + gen_python_files_in_dir(p, root, include_regex, exclude_regex, report) ) - elif p.is_file(): + elif p.is_file() or s == "-": # if a file was explicitly given, we don't care about its extension - sources.append(p) - elif s == "-": - sources.append(Path("-")) + sources.add(p) else: err(f"invalid path: {s}") - - if check and not diff: - write_back = WriteBack.NO - elif diff: - write_back = WriteBack.DIFF - else: - write_back = WriteBack.YES - mode = FileMode.AUTO_DETECT - if py36: - mode |= FileMode.PYTHON36 - if pyi: - mode |= FileMode.PYI - if skip_string_normalization: - mode |= FileMode.NO_STRING_NORMALIZATION - report = Report(check=check, quiet=quiet) if len(sources) == 0: - out("No paths given. Nothing to do 😴") + if verbose or not quiet: + out("No paths given. Nothing to do 😴") ctx.exit(0) - return - elif len(sources) == 1: + if len(sources) == 1: reformat_one( - src=sources[0], + src=sources.pop(), line_length=line_length, fast=fast, write_back=write_back, @@ -314,9 +356,10 @@ def main( ) finally: shutdown(loop) - if not quiet: - out("All done! ✨ 🍰 ✨") - click.echo(str(report)) + if verbose or not quiet: + bang = "💥 💔 💥" if report.return_code else "✨ 🍰 ✨" + out(f"All done! {bang}") + click.secho(str(report), err=True) ctx.exit(report.return_code) @@ -356,7 +399,9 @@ def reformat_one( mode=mode, ): changed = Changed.YES - if write_back == WriteBack.YES and changed is not Changed.NO: + if (write_back is WriteBack.YES and changed is not Changed.CACHED) or ( + write_back is WriteBack.CHECK and changed is Changed.NO + ): write_cache(cache, [src], line_length, mode) report.done(src, changed) except Exception as exc: @@ -364,7 +409,7 @@ def reformat_one( async def schedule_formatting( - sources: List[Path], + sources: Set[Path], line_length: int, fast: bool, write_back: WriteBack, @@ -384,52 +429,60 @@ async def schedule_formatting( if write_back != WriteBack.DIFF: cache = read_cache(line_length, mode) sources, cached = filter_cached(cache, sources) - for src in cached: + for src in sorted(cached): report.done(src, Changed.CACHED) + if not sources: + return + cancelled = [] - 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 = { - loop.run_in_executor( - executor, - format_file_in_place, - src, - line_length, - fast, - write_back, - mode, - lock, - ): src - for src in sorted(sources) - } - pending: Iterable[asyncio.Task] = tasks.keys() - try: - loop.add_signal_handler(signal.SIGINT, cancel, pending) - loop.add_signal_handler(signal.SIGTERM, cancel, pending) - except NotImplementedError: - # There are no good alternatives for these on Windows - pass - while pending: - done, _ = await asyncio.wait(pending, return_when=asyncio.FIRST_COMPLETED) - for task in done: - src = tasks.pop(task) - if 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) + sources_to_cache = [] + 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 = { + loop.run_in_executor( + executor, + format_file_in_place, + src, + line_length, + fast, + write_back, + mode, + lock, + ): src + for src in sorted(sources) + } + pending: Iterable[asyncio.Task] = tasks.keys() + try: + loop.add_signal_handler(signal.SIGINT, cancel, pending) + loop.add_signal_handler(signal.SIGTERM, cancel, pending) + except NotImplementedError: + # There are no good alternatives for these on Windows. + pass + while pending: + done, _ = await asyncio.wait(pending, return_when=asyncio.FIRST_COMPLETED) + for task in done: + src = tasks.pop(task) + if task.cancelled(): + cancelled.append(task) + elif task.exception(): + report.failed(src, str(task.exception())) + else: + changed = Changed.YES if task.result() else Changed.NO + # If the file was written back or was successfully checked as + # well-formatted, store this information in the cache. + if write_back is WriteBack.YES or ( + write_back is WriteBack.CHECK and changed is Changed.NO + ): + sources_to_cache.append(src) + report.done(src, changed) if cancelled: await asyncio.gather(*cancelled, loop=loop, return_exceptions=True) - if write_back == WriteBack.YES and formatted: - write_cache(cache, formatted, line_length, mode) + if sources_to_cache: + write_cache(cache, sources_to_cache, line_length, mode) def format_file_in_place( @@ -442,13 +495,16 @@ def format_file_in_place( ) -> bool: """Format file under `src` path. Return True if changed. - If `write_back` is True, write reformatted code back to stdout. + If `write_back` is DIFF, write a diff to stdout. If it is YES, write reformatted + code to the file. `line_length` and `fast` options are passed to :func:`format_file_contents`. """ if src.suffix == ".pyi": mode |= FileMode.PYI - with tokenize.open(src) as src_buffer: - src_contents = src_buffer.read() + + then = datetime.utcfromtimestamp(src.stat().st_mtime) + with open(src, "rb") as buf: + src_contents, encoding, newline = decode_bytes(buf.read()) try: dst_contents = format_file_contents( src_contents, line_length=line_length, fast=fast, mode=mode @@ -457,16 +513,24 @@ def format_file_in_place( return False if write_back == write_back.YES: - with open(src, "w", encoding=src_buffer.encoding) as f: + with open(src, "w", encoding=encoding, newline=newline) as f: f.write(dst_contents) elif write_back == write_back.DIFF: - src_name = f"{src} (original)" - dst_name = f"{src} (formatted)" + now = datetime.utcnow() + src_name = f"{src}\t{then} +0000" + dst_name = f"{src}\t{now} +0000" diff_contents = diff(src_contents, dst_contents, src_name, dst_name) if lock: lock.acquire() try: - sys.stdout.write(diff_contents) + f = io.TextIOWrapper( + sys.stdout.buffer, + encoding=encoding, + newline=newline, + write_through=True, + ) + f.write(diff_contents) + f.detach() finally: if lock: lock.release() @@ -481,11 +545,13 @@ def format_stdin_to_stdout( ) -> bool: """Format file on stdin. Return True if changed. - If `write_back` is True, write reformatted code back to stdout. + If `write_back` is YES, write reformatted code back to stdout. If it is DIFF, + write a diff to stdout. `line_length`, `fast`, `is_pyi`, and `force_py36` arguments are passed to :func:`format_file_contents`. """ - src = sys.stdin.read() + then = datetime.utcnow() + src, encoding, newline = decode_bytes(sys.stdin.buffer.read()) dst = src try: dst = format_file_contents(src, line_length=line_length, fast=fast, mode=mode) @@ -495,12 +561,17 @@ def format_stdin_to_stdout( return False finally: + f = io.TextIOWrapper( + sys.stdout.buffer, encoding=encoding, newline=newline, write_through=True + ) if write_back == WriteBack.YES: - sys.stdout.write(dst) + f.write(dst) elif write_back == WriteBack.DIFF: - src_name = " (original)" - dst_name = " (formatted)" - sys.stdout.write(diff(src, dst, src_name, dst_name)) + now = datetime.utcnow() + src_name = f"STDIN\t{then} +0000" + dst_name = f"STDOUT\t{now} +0000" + f.write(diff(src, dst, src_name, dst_name)) + f.detach() def format_file_contents( @@ -542,10 +613,12 @@ def format_str( is_pyi = bool(mode & FileMode.PYI) py36 = bool(mode & FileMode.PYTHON36) or is_python36(src_node) normalize_strings = not bool(mode & FileMode.NO_STRING_NORMALIZATION) + normalize_fmt_off(src_node) lines = LineGenerator( remove_u_prefix=py36 or "unicode_literals" in future_imports, is_pyi=is_pyi, normalize_strings=normalize_strings, + allow_underscores=py36, ) elt = EmptyLineTracker(is_pyi=is_pyi) empty_line = Line() @@ -561,6 +634,23 @@ def format_str( return dst_contents +def decode_bytes(src: bytes) -> Tuple[FileContent, Encoding, NewLine]: + """Return a tuple of (decoded_contents, encoding, newline). + + `newline` is either CRLF or LF but `decoded_contents` is decoded with + universal newlines (i.e. only contains LF). + """ + srcbuf = io.BytesIO(src) + encoding, lines = tokenize.detect_encoding(srcbuf.readline) + if not lines: + return "", encoding, "\n" + + newline = "\r\n" if b"\r\n" == lines[0][-2:] else "\n" + srcbuf.seek(0) + with io.TextIOWrapper(srcbuf, encoding) as tiow: + return tiow.read(), encoding, newline + + GRAMMARS = [ pygram.python_grammar_no_print_statement_no_exec_statement, pygram.python_grammar_no_print_statement, @@ -571,9 +661,8 @@ GRAMMARS = [ def lib2to3_parse(src_txt: str) -> Node: """Given a string with source, return the lib2to3 Node.""" grammar = pygram.python_grammar_no_print_statement - if src_txt[-1] != "\n": - nl = "\r\n" if "\r\n" in src_txt[:1024] else "\n" - src_txt += nl + if src_txt[-1:] != "\n": + src_txt += "\n" for grammar in GRAMMARS: drv = driver.Driver(grammar, pytree.convert) try: @@ -656,13 +745,15 @@ class DebugVisitor(Visitor[T]): out(f" {node.value!r}", fg="blue", bold=False) @classmethod - def show(cls, code: str) -> None: + def show(cls, code: Union[str, Leaf, Node]) -> None: """Pretty-print the lib2to3 AST of a given string of `code`. Convenience method for debugging. """ v: DebugVisitor[None] = DebugVisitor() - list(v.visit(lib2to3_parse(code))) + if isinstance(code, str): + code = lib2to3_parse(code) + list(v.visit(code)) KEYWORDS = set(keyword.kwlist) @@ -679,6 +770,7 @@ STATEMENT = { syms.classdef, } STANDALONE_COMMENT = 153 +token.tok_name[STANDALONE_COMMENT] = "STANDALONE_COMMENT" LOGIC_OPERATORS = {"and", "or"} COMPARATORS = { token.LESS, @@ -717,6 +809,7 @@ UNPACKING_PARENTS = { syms.dictsetmaker, syms.listmaker, syms.testlist_gexp, + syms.testlist_star_expr, } TEST_DESCENDANTS = { syms.test, @@ -1039,6 +1132,13 @@ class Line: return False + def contains_multiline_strings(self) -> bool: + for leaf in self.leaves: + if is_multiline_string(leaf): + return True + + return False + def maybe_remove_trailing_comma(self, closing: Leaf) -> bool: """Remove trailing comma if there is one and it's safe.""" if not ( @@ -1064,7 +1164,7 @@ class Line: self.remove_trailing_comma() return True - # Otheriwsse, if the trailing one is the only one, we might mistakenly + # Otherwise, 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 commas = 0 @@ -1120,6 +1220,9 @@ class Line: Provide a non-negative leaf `_index` to speed up the function. """ + if not self.comments: + return + if _index == -1: for _index, _leaf in enumerate(self.leaves): if leaf is _leaf: @@ -1143,18 +1246,18 @@ class Line: def is_complex_subscript(self, leaf: Leaf) -> bool: """Return True iff `leaf` is part of a slice with non-trivial exprs.""" - open_lsqb = ( - leaf if leaf.type == token.LSQB else self.bracket_tracker.get_open_lsqb() - ) + open_lsqb = self.bracket_tracker.get_open_lsqb() if open_lsqb is None: return False subscript_start = open_lsqb.next_sibling - if ( - isinstance(subscript_start, Node) - and subscript_start.type == syms.subscriptlist - ): - subscript_start = child_towards(subscript_start, leaf) + + if isinstance(subscript_start, Node): + if subscript_start.type == syms.listmaker: + return False + + if subscript_start.type == syms.subscriptlist: + subscript_start = child_towards(subscript_start, leaf) return subscript_start is not None and any( n.type in TEST_DESCENDANTS for n in subscript_start.pre_order() ) @@ -1179,55 +1282,6 @@ class Line: return bool(self.leaves or self.comments) -class UnformattedLines(Line): - """Just like :class:`Line` but stores lines which aren't reformatted.""" - - def append(self, leaf: Leaf, preformatted: bool = True) -> None: - """Just add a new `leaf` to the end of the lines. - - The `preformatted` argument is ignored. - - Keeps track of indentation `depth`, which is useful when the user - says `# fmt: on`. Otherwise, doesn't do anything with the `leaf`. - """ - try: - list(generate_comments(leaf)) - except FormatOn as f_on: - self.leaves.append(f_on.leaf_from_consumed(leaf)) - raise - - self.leaves.append(leaf) - if leaf.type == token.INDENT: - self.depth += 1 - elif leaf.type == token.DEDENT: - self.depth -= 1 - - def __str__(self) -> str: - """Render unformatted lines from leaves which were added with `append()`. - - `depth` is not used for indentation in this case. - """ - if not self: - return "\n" - - res = "" - for leaf in self.leaves: - res += str(leaf) - return res - - def append_comment(self, comment: Leaf) -> bool: - """Not implemented in this class. Raises `NotImplementedError`.""" - raise NotImplementedError("Unformatted lines don't store comments separately.") - - def maybe_remove_trailing_comma(self, closing: Leaf) -> bool: - """Does nothing and returns False.""" - return False - - def maybe_increment_for_loop_variable(self, leaf: Leaf) -> bool: - """Does nothing and returns False.""" - return False - - @dataclass class EmptyLineTracker: """Provides a stateful method that returns the number of potential extra @@ -1249,9 +1303,6 @@ class EmptyLineTracker: This is for separating `def`, `async def` and `class` with extra empty lines (two on module-level). """ - if isinstance(current_line, UnformattedLines): - return 0, 0 - before, after = self._maybe_empty_lines(current_line) before -= self.previous_after self.previous_after = after @@ -1277,44 +1328,8 @@ class EmptyLineTracker: before = 0 if depth else 1 else: 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: - self.previous_defs.append(depth) - if self.previous_line is None: - # Don't insert empty lines before the first line in the file. - return 0, 0 - - if self.previous_line.is_decorator: - return 0, 0 - - if self.previous_line.depth < current_line.depth and ( - self.previous_line.is_class or self.previous_line.is_def - ): - return 0, 0 - - if ( - self.previous_line.is_comment - and self.previous_line.depth == current_line.depth - and before == 0 - ): - return 0, 0 - - if self.is_pyi: - if self.previous_line.depth > current_line.depth: - newlines = 1 - elif current_line.is_class or self.previous_line.is_class: - if current_line.is_stub_class and self.previous_line.is_stub_class: - newlines = 0 - else: - newlines = 1 - else: - newlines = 0 - else: - newlines = 2 - if current_line.depth and newlines: - newlines -= 1 - return newlines, 0 + if current_line.is_decorator or current_line.is_def or current_line.is_class: + return self._maybe_empty_lines_for_class_or_def(current_line, before) if ( self.previous_line @@ -1333,6 +1348,50 @@ class EmptyLineTracker: return before, 0 + def _maybe_empty_lines_for_class_or_def( + self, current_line: Line, before: int + ) -> Tuple[int, int]: + if not current_line.is_decorator: + self.previous_defs.append(current_line.depth) + if self.previous_line is None: + # Don't insert empty lines before the first line in the file. + return 0, 0 + + if self.previous_line.is_decorator: + return 0, 0 + + if self.previous_line.depth < current_line.depth and ( + self.previous_line.is_class or self.previous_line.is_def + ): + return 0, 0 + + if ( + self.previous_line.is_comment + and self.previous_line.depth == current_line.depth + and before == 0 + ): + return 0, 0 + + if self.is_pyi: + if self.previous_line.depth > current_line.depth: + newlines = 1 + elif current_line.is_class or self.previous_line.is_class: + if current_line.is_stub_class and self.previous_line.is_stub_class: + # No blank line between classes with an empty body + newlines = 0 + else: + newlines = 1 + elif current_line.is_def and not self.previous_line.is_def: + # Blank line between a block of functions and a block of non-functions + newlines = 1 + else: + newlines = 0 + else: + newlines = 2 + if current_line.depth and newlines: + newlines -= 1 + return newlines, 0 + @dataclass class LineGenerator(Visitor[Line]): @@ -1346,8 +1405,9 @@ class LineGenerator(Visitor[Line]): normalize_strings: bool = True current_line: Line = Factory(Line) remove_u_prefix: bool = False + allow_underscores: bool = False - def line(self, indent: int = 0, type: Type[Line] = Line) -> Iterator[Line]: + def line(self, indent: int = 0) -> Iterator[Line]: """Generate a line. If the line is empty, only emit if it makes sense. @@ -1356,67 +1416,41 @@ class LineGenerator(Visitor[Line]): If any lines were generated, set up a new current_line. """ if not self.current_line: - if self.current_line.__class__ == type: - self.current_line.depth += indent - else: - self.current_line = type(depth=self.current_line.depth + indent) + 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 = type(depth=complete_line.depth + indent) + self.current_line = Line(depth=complete_line.depth + indent) yield complete_line - def visit(self, node: LN) -> Iterator[Line]: - """Main method to visit `node` and its children. - - Yields :class:`Line` objects. - """ - if isinstance(self.current_line, UnformattedLines): - # File contained `# fmt: off` - yield from self.visit_unformatted(node) - - else: - yield from super().visit(node) - def visit_default(self, node: LN) -> Iterator[Line]: """Default `visit_*()` implementation. Recurses to children of `node`.""" if isinstance(node, Leaf): 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) + 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: - normalize_prefix(node, inside_brackets=any_open_brackets) - if self.normalize_strings and node.type == token.STRING: - normalize_string_prefix(node, remove_u_prefix=self.remove_u_prefix) - normalize_string_quotes(node) - if node.type not in WHITESPACE: - self.current_line.append(node) + else: + # regular standalone comment + yield from self.line() + + self.current_line.append(comment) + yield from self.line() + + normalize_prefix(node, inside_brackets=any_open_brackets) + if self.normalize_strings and node.type == token.STRING: + normalize_string_prefix(node, remove_u_prefix=self.remove_u_prefix) + normalize_string_quotes(node) + if node.type == token.NUMBER: + normalize_numeric_literal(node, self.allow_underscores) + if node.type not in WHITESPACE: + self.current_line.append(node) yield from super().visit_default(node) def visit_INDENT(self, node: Node) -> Iterator[Line]: @@ -1513,23 +1547,10 @@ class LineGenerator(Visitor[Line]): yield from self.visit_default(leaf) yield from self.line() - def visit_unformatted(self, node: LN) -> Iterator[Line]: - """Used when file contained a `# fmt: off`.""" - if isinstance(node, Node): - for child in node.children: - yield from self.visit(child) - - else: - try: - self.current_line.append(node) - except FormatOn as f_on: - f_on.trim_prefix(node) - yield from self.line() - yield from self.visit(node) - - if node.type == token.ENDMARKER: - # somebody decided not to put a final `# fmt: on` - yield from self.line() + def visit_STANDALONE_COMMENT(self, leaf: Leaf) -> Iterator[Line]: + if not self.current_line.bracket_tracker.any_open_brackets(): + yield from self.line() + yield from self.visit_default(leaf) def __attrs_post_init__(self) -> None: """You are in a twisty little maze of passages.""" @@ -1772,7 +1793,7 @@ def whitespace(leaf: Leaf, *, complex_subscript: bool) -> str: # noqa C901 elif prevp.type == token.EQUAL and prevp_parent.type == syms.argument: return NO - elif t == token.NAME or t == token.NUMBER: + elif t in {token.NAME, token.NUMBER, token.STRING}: return NO elif p.type == syms.import_from: @@ -1819,6 +1840,31 @@ def child_towards(ancestor: Node, descendant: LN) -> Optional[LN]: return node +def container_of(leaf: Leaf) -> LN: + """Return `leaf` or one of its ancestors that is the topmost container of it. + + By "container" we mean a node where `leaf` is the very first child. + """ + same_prefix = leaf.prefix + container: LN = leaf + while container: + parent = container.parent + if parent is None: + break + + if parent.children[0].prefix != same_prefix: + break + + if parent.type == syms.file_input: + break + + if parent.prev_sibling is not None and parent.prev_sibling.type in BRACKETS: + break + + container = parent + return container + + def is_split_after_delimiter(leaf: Leaf, previous: Leaf = None) -> int: """Return the priority of the `leaf` delimiter, given a line break after it. @@ -1924,6 +1970,10 @@ def is_split_before_delimiter(leaf: Leaf, previous: Leaf = None) -> int: return 0 +FMT_OFF = {"# fmt: off", "# fmt:off", "# yapf: disable"} +FMT_ON = {"# fmt: on", "# fmt:on", "# yapf: enable"} + + def generate_comments(leaf: LN) -> Iterator[Leaf]: """Clean the prefix of the `leaf` and generate comments from it, if any. @@ -1943,16 +1993,27 @@ def generate_comments(leaf: LN) -> Iterator[Leaf]: Inline comments are emitted as regular token.COMMENT leaves. Standalone are emitted with a fake STANDALONE_COMMENT token identifier. """ - p = leaf.prefix - if not p: - return + for pc in list_comments(leaf.prefix, is_endmarker=leaf.type == token.ENDMARKER): + yield Leaf(pc.type, pc.value, prefix="\n" * pc.newlines) - if "#" not in p: - return + +@dataclass +class ProtoComment: + type: int # token.COMMENT or STANDALONE_COMMENT + value: str # content of the comment + newlines: int # how many newlines before the comment + consumed: int # how many characters of the original leaf's prefix did we consume + + +@lru_cache(maxsize=4096) +def list_comments(prefix: str, *, is_endmarker: bool) -> List[ProtoComment]: + result: List[ProtoComment] = [] + if not prefix or "#" not in prefix: + return result consumed = 0 nlines = 0 - for index, line in enumerate(p.split("\n")): + for index, line in enumerate(prefix.split("\n")): consumed += len(line) + 1 # adding the length of the split '\n' line = line.lstrip() if not line: @@ -1960,25 +2021,18 @@ def generate_comments(leaf: LN) -> Iterator[Leaf]: if not line.startswith("#"): continue - if index == 0 and leaf.type != token.ENDMARKER: + if index == 0 and not is_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"}: - 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) - + result.append( + ProtoComment( + type=comment_type, value=comment, newlines=nlines, consumed=consumed + ) + ) nlines = 0 + return result def make_comment(content: str) -> str: @@ -2013,7 +2067,7 @@ def split_line( If `py36` is True, splitting may generate syntax that is only compatible with Python 3.6 and later. """ - if isinstance(line, UnformattedLines) or line.is_comment: + if line.is_comment: yield line return @@ -2158,32 +2212,50 @@ def right_hand_split( 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 + body.should_explode = should_explode(body, opening_bracket) + bracket_split_succeeded_or_raise(head, body, tail) if ( + # the body shouldn't be exploded + not body.should_explode # the opening bracket is an optional paren - opening_bracket.type == token.LPAR + and opening_bracket.type == token.LPAR and not opening_bracket.value # the closing bracket is an optional paren and closing_bracket.type == token.RPAR and not closing_bracket.value - # there are no standalone comments in the body - and not line.contains_standalone_comments(0) - # and it's not an import (optional parens are the only thing we can split - # on in this case; attempting a split without them is a waste of time) + # it's not an import (optional parens are the only thing we can split on + # in this case; attempting a split without them is a waste of time) and not line.is_import + # there are no standalone comments in the body + and not body.contains_standalone_comments(0) + # and we can actually remove the parens + and can_omit_invisible_parens(body, line_length) ): omit = {id(closing_bracket), *omit} - if can_omit_invisible_parens(body, line_length): - try: - yield from right_hand_split(line, line_length, py36=py36, omit=omit) - return - except CannotSplit: - pass + try: + yield from right_hand_split(line, line_length, py36=py36, omit=omit) + return + + except CannotSplit: + if not ( + can_be_split(body) + or is_line_short_enough(body, line_length=line_length) + ): + raise CannotSplit( + "Splitting failed, body is still too long and can't be split." + ) + + elif head.contains_multiline_strings() or tail.contains_multiline_strings(): + raise CannotSplit( + "The current optional pair of parentheses is bound to fail to " + "satisfy the splitting algorithm because the head or the tail " + "contains multiline strings which by definition never fit one " + "line." + ) ensure_visible(opening_bracket) ensure_visible(closing_bracket) - body.should_explode = should_explode(body, opening_bracket) for result in (head, body, tail): if result: yield result @@ -2398,8 +2470,8 @@ def normalize_string_quotes(leaf: Leaf) -> None: 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}") + 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)] if "r" in prefix.casefold(): if unescaped_new_quote.search(body): @@ -2410,15 +2482,21 @@ def normalize_string_quotes(leaf: Leaf) -> None: # Do not introduce or remove backslashes in raw strings new_body = body else: - # remove unnecessary quotes + # remove unnecessary escapes 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 + # Consider the string without unnecessary escapes 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] == '"': + if "f" in prefix.casefold(): + matches = re.findall(r"[^{]\{(.*?)\}[^}]", new_body) + for m in matches: + if "\\" in str(m): + # Do not introduce backslashes in interpolated expressions + return + if new_quote == '"""' and new_body[-1:] == '"': # edge case: new_body = new_body[:-1] + '\\"' orig_escape_count = body.count("\\") @@ -2432,6 +2510,63 @@ def normalize_string_quotes(leaf: Leaf) -> None: leaf.value = f"{prefix}{new_quote}{new_body}{new_quote}" +def normalize_numeric_literal(leaf: Leaf, allow_underscores: bool) -> None: + """Normalizes numeric (float, int, and complex) literals. + + All letters used in the representation are normalized to lowercase, long number + literals are split using underscores. + """ + text = leaf.value.lower() + if text.startswith(("0o", "0x", "0b")): + # Leave octal, hex, and binary literals alone. + pass + elif "e" in text: + before, after = text.split("e") + sign = "" + if after.startswith("-"): + after = after[1:] + sign = "-" + elif after.startswith("+"): + after = after[1:] + before = format_float_or_int_string(before, allow_underscores) + after = format_int_string(after, allow_underscores) + text = f"{before}e{sign}{after}" + elif text.endswith(("j", "l")): + number = text[:-1] + suffix = text[-1] + text = f"{format_float_or_int_string(number, allow_underscores)}{suffix}" + else: + text = format_float_or_int_string(text, allow_underscores) + leaf.value = text + + +def format_float_or_int_string(text: str, allow_underscores: bool) -> str: + """Formats a float string like "1.0".""" + if "." not in text: + return format_int_string(text, allow_underscores) + + before, after = text.split(".") + before = format_int_string(before, allow_underscores) if before else "0" + after = format_int_string(after, allow_underscores) if after else "0" + return f"{before}.{after}" + + +def format_int_string(text: str, allow_underscores: bool) -> str: + """Normalizes underscores in a string to e.g. 1_000_000. + + Input must be a string of at least six digits and optional underscores. + """ + if not allow_underscores: + return text + + text = text.replace("_", "") + if len(text) <= 6: + # No underscores for numbers <= 6 digits long. + return text + + return format(int(text), "3_") + + def normalize_invisible_parens(node: Node, parens_after: Set[str]) -> None: """Make existing optional parentheses invisible or create new ones. @@ -2441,10 +2576,10 @@ def normalize_invisible_parens(node: Node, parens_after: Set[str]) -> None: Standardizes on visible parentheses for single-element tuples, and keeps existing visible parentheses for other tuples and generator expressions. """ - try: - list(generate_comments(node)) - except FormatOff: - return # This `node` has a prefix with `# fmt: off`, don't mess with parens. + for pc in list_comments(node.prefix, is_endmarker=False): + if pc.value in FMT_OFF: + # This `node` has a prefix with `# fmt: off`, don't mess with parens. + return check_lpar = False for index, child in enumerate(list(node.children)): @@ -2480,8 +2615,85 @@ def normalize_invisible_parens(node: Node, parens_after: Set[str]) -> None: check_lpar = isinstance(child, Leaf) and child.value in parens_after +def normalize_fmt_off(node: Node) -> None: + """Convert content between `# fmt: off`/`# fmt: on` into standalone comments.""" + try_again = True + while try_again: + try_again = convert_one_fmt_off_pair(node) + + +def convert_one_fmt_off_pair(node: Node) -> bool: + """Convert content of a single `# fmt: off`/`# fmt: on` into a standalone comment. + + Returns True if a pair was converted. + """ + for leaf in node.leaves(): + previous_consumed = 0 + for comment in list_comments(leaf.prefix, is_endmarker=False): + if comment.value in FMT_OFF: + # We only want standalone comments. If there's no previous leaf or + # the previous leaf is indentation, it's a standalone comment in + # disguise. + if comment.type != STANDALONE_COMMENT: + prev = preceding_leaf(leaf) + if prev and prev.type not in WHITESPACE: + continue + + ignored_nodes = list(generate_ignored_nodes(leaf)) + if not ignored_nodes: + continue + + first = ignored_nodes[0] # Can be a container node with the `leaf`. + parent = first.parent + prefix = first.prefix + first.prefix = prefix[comment.consumed :] + hidden_value = ( + comment.value + "\n" + "".join(str(n) for n in ignored_nodes) + ) + if hidden_value.endswith("\n"): + # That happens when one of the `ignored_nodes` ended with a NEWLINE + # leaf (possibly followed by a DEDENT). + hidden_value = hidden_value[:-1] + first_idx = None + for ignored in ignored_nodes: + index = ignored.remove() + if first_idx is None: + first_idx = index + assert parent is not None, "INTERNAL ERROR: fmt: on/off handling (1)" + assert first_idx is not None, "INTERNAL ERROR: fmt: on/off handling (2)" + parent.insert_child( + first_idx, + Leaf( + STANDALONE_COMMENT, + hidden_value, + prefix=prefix[:previous_consumed] + "\n" * comment.newlines, + ), + ) + return True + + previous_consumed = comment.consumed + + return False + + +def generate_ignored_nodes(leaf: Leaf) -> Iterator[LN]: + """Starting from the container of `leaf`, generate all leaves until `# fmt: on`. + + Stops at the end of the block. + """ + container: Optional[LN] = container_of(leaf) + while container is not None and container.type != token.ENDMARKER: + for comment in list_comments(container.prefix, is_endmarker=False): + if comment.value in FMT_ON: + return + + yield container + + container = container.next_sibling + + def maybe_make_parens_invisible_in_atom(node: LN) -> bool: - """If it's safe, make the parens in the atom `node` invisible, recusively.""" + """If it's safe, make the parens in the atom `node` invisible, recursively.""" if ( node.type != syms.atom or is_empty_tuple(node) @@ -2761,7 +2973,23 @@ def generate_trailers_to_omit(line: Line, line_length: int) -> Iterator[Set[Leaf def get_future_imports(node: Node) -> Set[str]: """Return a set of __future__ imports in the file.""" - imports = set() + imports: Set[str] = set() + + def get_imports_from_children(children: List[LN]) -> Generator[str, None, None]: + for child in children: + if isinstance(child, Leaf): + if child.type == token.NAME: + yield child.value + elif child.type == syms.import_as_name: + orig_name = child.children[0] + assert isinstance(orig_name, Leaf), "Invalid syntax parsing imports" + assert orig_name.type == token.NAME, "Invalid syntax parsing imports" + yield orig_name.value + elif child.type == syms.import_as_names: + yield from get_imports_from_children(child.children) + else: + assert False, "Invalid syntax parsing imports" + for child in node.children: if child.type != syms.simple_stmt: break @@ -2780,37 +3008,48 @@ def get_future_imports(node: Node) -> Set[str]: module_name = first_child.children[1] if not isinstance(module_name, Leaf) or module_name.value != "__future__": break - for import_from_child in first_child.children[3:]: - if isinstance(import_from_child, Leaf): - if import_from_child.type == token.NAME: - imports.add(import_from_child.value) - else: - assert import_from_child.type == syms.import_as_names - for leaf in import_from_child.children: - if isinstance(leaf, Leaf) and leaf.type == token.NAME: - imports.add(leaf.value) + imports |= set(get_imports_from_children(first_child.children[3:])) else: break return imports def gen_python_files_in_dir( - path: Path, root: Path, include: Pattern[str], exclude: Pattern[str] + path: Path, + root: Path, + include: Pattern[str], + exclude: Pattern[str], + report: "Report", ) -> Iterator[Path]: """Generate all files under `path` whose paths are not excluded by the `exclude` regex, but are included by the `include` regex. + + Symbolic links pointing outside of the `root` directory are ignored. + + `report` is where output about exclusions goes. """ assert root.is_absolute(), f"INTERNAL ERROR: `root` must be absolute but is {root}" for child in path.iterdir(): - normalized_path = child.resolve().relative_to(root).as_posix() + try: + normalized_path = "/" + child.resolve().relative_to(root).as_posix() + except ValueError: + if child.is_symlink(): + report.path_ignored( + child, f"is a symbolic link that points outside {root}" + ) + continue + + raise + if child.is_dir(): normalized_path += "/" exclude_match = exclude.search(normalized_path) if exclude_match and exclude_match.group(0): + report.path_ignored(child, f"matches the --exclude regular expression") continue if child.is_dir(): - yield from gen_python_files_in_dir(child, root, include, exclude) + yield from gen_python_files_in_dir(child, root, include, exclude, report) elif child.is_file(): include_match = include.search(normalized_path) @@ -2818,7 +3057,8 @@ def gen_python_files_in_dir( yield child -def find_project_root(srcs: List[str]) -> Path: +@lru_cache() +def find_project_root(srcs: Iterable[str]) -> Path: """Return a directory containing .git, .hg, or pyproject.toml. That directory can be one of the directories passed in `srcs` or their @@ -2853,6 +3093,7 @@ class Report: check: bool = False quiet: bool = False + verbose: bool = False change_count: int = 0 same_count: int = 0 failure_count: int = 0 @@ -2861,11 +3102,11 @@ class Report: """Increment the counter for successful reformatting. Write out a message.""" if changed is Changed.YES: reformatted = "would reformat" if self.check else "reformatted" - if not self.quiet: + if self.verbose or not self.quiet: out(f"{reformatted} {src}") self.change_count += 1 else: - if not self.quiet: + if self.verbose: if changed is Changed.NO: msg = f"{src} already well formatted, good job." else: @@ -2878,6 +3119,10 @@ class Report: err(f"error: cannot format {src}: {message}") self.failure_count += 1 + def path_ignored(self, path: Path, message: str) -> None: + if self.verbose: + out(f"{path} ignored: {message}", bold=False) + @property def return_code(self) -> int: """Return the exit code that the app should use. @@ -2888,7 +3133,7 @@ class Report: - otherwise return 0. """ # According to http://tldp.org/LDP/abs/html/exitcodes.html starting with - # 126 we have special returncodes reserved by the shell. + # 126 we have special return codes reserved by the shell. if self.failure_count: return 123 @@ -3071,6 +3316,16 @@ def sub_twice(regex: Pattern[str], replacement: str, original: str) -> str: return regex.sub(replacement, regex.sub(replacement, original)) +def re_compile_maybe_verbose(regex: str) -> Pattern[str]: + """Compile a regular expression string in `regex`. + + If it contains newlines, use verbose mode. + """ + if "\n" in regex: + regex = "(?x)" + regex + return re.compile(regex) + + def enumerate_reversed(sequence: Sequence[T]) -> Iterator[Tuple[Index, T]]: """Like `reversed(enumerate(sequence))` if that were possible.""" index = len(sequence) - 1 @@ -3116,6 +3371,42 @@ def is_line_short_enough(line: Line, *, line_length: int, line_str: str = "") -> ) +def can_be_split(line: Line) -> bool: + """Return False if the line cannot be split *for sure*. + + This is not an exhaustive search but a cheap heuristic that we can use to + avoid some unfortunate formattings (mostly around wrapping unsplittable code + in unnecessary parentheses). + """ + leaves = line.leaves + if len(leaves) < 2: + return False + + if leaves[0].type == token.STRING and leaves[1].type == token.DOT: + call_count = 0 + dot_count = 0 + next = leaves[-1] + for leaf in leaves[-2::-1]: + if leaf.type in OPENING_BRACKETS: + if next.type not in CLOSING_BRACKETS: + return False + + call_count += 1 + elif leaf.type == token.DOT: + dot_count += 1 + elif leaf.type == token.NAME: + if not (next.type == token.DOT or next.type in OPENING_BRACKETS): + return False + + elif leaf.type not in CLOSING_BRACKETS: + return False + + if dot_count > 1 and call_count > 1: + return False + + return True + + def can_omit_invisible_parens(line: Line, line_length: int) -> bool: """Does `line` have a shape safe to reformat without optional parens around it? @@ -3206,12 +3497,7 @@ def can_omit_invisible_parens(line: Line, line_length: int) -> bool: def get_cache_file(line_length: int, mode: FileMode) -> Path: - pyi = bool(mode & FileMode.PYI) - py36 = bool(mode & FileMode.PYTHON36) - return ( - CACHE_DIR - / f"cache.{line_length}{'.pyi' if pyi else ''}{'.py36' if py36 else ''}.pickle" - ) + return CACHE_DIR / f"cache.{line_length}.{mode.value}.pickle" def read_cache(line_length: int, mode: FileMode) -> Cache: @@ -3238,26 +3524,24 @@ def get_cache_info(path: Path) -> CacheInfo: 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. +def filter_cached(cache: Cache, sources: Iterable[Path]) -> Tuple[Set[Path], Set[Path]]: + """Split an iterable of paths in `sources` into two sets. - 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. + The first contains paths of files that modified on disk or are not in the + cache. The other contains paths to non-modified files. """ - todo, done = [], [] + todo, done = set(), set() for src in sources: src = src.resolve() if cache.get(src) != get_cache_info(src): - todo.append(src) + todo.add(src) else: - done.append(src) + done.add(src) return todo, done def write_cache( - cache: Cache, sources: List[Path], line_length: int, mode: FileMode + cache: Cache, sources: Iterable[Path], line_length: int, mode: FileMode ) -> None: """Update the cache file.""" cache_file = get_cache_file(line_length, mode) @@ -3271,5 +3555,28 @@ def write_cache( pass +def patch_click() -> None: + """Make Click not crash. + + On certain misconfigured environments, Python 3 selects the ASCII encoding as the + default which restricts paths that it can access during the lifetime of the + application. Click refuses to work in this scenario by raising a RuntimeError. + + In case of Black the likelihood that non-ASCII characters are going to be used in + file paths is minimal since it's Python source code. Moreover, this crash was + spurious on Python 3.7 thanks to PEP 538 and PEP 540. + """ + try: + from click import core + from click import _unicodefun # type: ignore + except ModuleNotFoundError: + return + + for module in (core, _unicodefun): + if hasattr(module, "_verify_python3_env"): + module._verify_python3_env = lambda: None + + if __name__ == "__main__": + patch_click() main()