X-Git-Url: https://git.madduck.net/etc/vim.git/blobdiff_plain/4a953b7241ce5f8bcac985fa33fdf3af4f42c0de..5206560946f9941f95c2e96fa994485148c28a3a:/black.py?ds=sidebyside diff --git a/black.py b/black.py index 90aaf38..ce27653 100644 --- a/black.py +++ b/black.py @@ -16,6 +16,7 @@ import signal import sys import tempfile import tokenize +import traceback from typing import ( Any, Callable, @@ -40,6 +41,7 @@ from appdirs import user_cache_dir from attr import dataclass, evolve, Factory import click import toml +from typed_ast import ast3, ast27 # lib2to3 fork from blib2to3.pytree import Node, Leaf, type_repr @@ -135,19 +137,28 @@ class Feature(Enum): NUMERIC_UNDERSCORES = 3 TRAILING_COMMA_IN_CALL = 4 TRAILING_COMMA_IN_DEF = 5 + # The following two feature-flags are mutually exclusive, and exactly one should be + # set for every version of python. + ASYNC_IDENTIFIERS = 6 + ASYNC_KEYWORDS = 7 VERSION_TO_FEATURES: Dict[TargetVersion, Set[Feature]] = { - TargetVersion.PY27: set(), - TargetVersion.PY33: {Feature.UNICODE_LITERALS}, - TargetVersion.PY34: {Feature.UNICODE_LITERALS}, - TargetVersion.PY35: {Feature.UNICODE_LITERALS, Feature.TRAILING_COMMA_IN_CALL}, + TargetVersion.PY27: {Feature.ASYNC_IDENTIFIERS}, + TargetVersion.PY33: {Feature.UNICODE_LITERALS, Feature.ASYNC_IDENTIFIERS}, + TargetVersion.PY34: {Feature.UNICODE_LITERALS, Feature.ASYNC_IDENTIFIERS}, + TargetVersion.PY35: { + Feature.UNICODE_LITERALS, + Feature.TRAILING_COMMA_IN_CALL, + Feature.ASYNC_IDENTIFIERS, + }, TargetVersion.PY36: { Feature.UNICODE_LITERALS, Feature.F_STRINGS, Feature.NUMERIC_UNDERSCORES, Feature.TRAILING_COMMA_IN_CALL, Feature.TRAILING_COMMA_IN_DEF, + Feature.ASYNC_IDENTIFIERS, }, TargetVersion.PY37: { Feature.UNICODE_LITERALS, @@ -155,6 +166,7 @@ VERSION_TO_FEATURES: Dict[TargetVersion, Set[Feature]] = { Feature.NUMERIC_UNDERSCORES, Feature.TRAILING_COMMA_IN_CALL, Feature.TRAILING_COMMA_IN_DEF, + Feature.ASYNC_KEYWORDS, }, TargetVersion.PY38: { Feature.UNICODE_LITERALS, @@ -162,6 +174,7 @@ VERSION_TO_FEATURES: Dict[TargetVersion, Set[Feature]] = { Feature.NUMERIC_UNDERSCORES, Feature.TRAILING_COMMA_IN_CALL, Feature.TRAILING_COMMA_IN_DEF, + Feature.ASYNC_KEYWORDS, }, } @@ -231,6 +244,7 @@ def read_pyproject_toml( @click.command(context_settings=dict(help_option_names=["-h", "--help"])) +@click.option("-c", "--code", type=str, help="Format the code passed in as a string.") @click.option( "-l", "--line-length", @@ -323,7 +337,7 @@ def read_pyproject_toml( "--quiet", is_flag=True, help=( - "Don't emit non-error messages to stderr. Errors are still emitted, " + "Don't emit non-error messages to stderr. Errors are still emitted; " "silence those with 2>/dev/null." ), ) @@ -357,6 +371,7 @@ def read_pyproject_toml( @click.pass_context def main( ctx: click.Context, + code: Optional[str], line_length: int, target_version: List[TargetVersion], check: bool, @@ -397,6 +412,9 @@ def main( ) if config and verbose: out(f"Using configuration from {config}.", bold=False, fg="blue") + if code is not None: + print(format_str(code, mode=mode)) + ctx.exit(0) try: include_regex = re_compile_maybe_verbose(include) except re.error: @@ -435,25 +453,12 @@ def main( report=report, ) else: - loop = asyncio.get_event_loop() - executor = ProcessPoolExecutor(max_workers=os.cpu_count()) - try: - loop.run_until_complete( - schedule_formatting( - sources=sources, - fast=fast, - write_back=write_back, - mode=mode, - report=report, - loop=loop, - executor=executor, - ) - ) - finally: - shutdown(loop) + reformat_many( + sources=sources, fast=fast, write_back=write_back, mode=mode, report=report + ) + if verbose or not quiet: - bang = "💥 💔 💥" if report.return_code else "✨ 🍰 ✨" - out(f"All done! {bang}") + out("Oh no! 💥 💔 💥" if report.return_code else "All done! ✨ 🍰 ✨") click.secho(str(report), err=True) ctx.exit(report.return_code) @@ -463,8 +468,7 @@ def reformat_one( ) -> None: """Reformat a single file under `src` without spawning child processes. - If `quiet` is True, non-error messages are not output. `line_length`, - `write_back`, `fast` and `pyi` options are passed to + `fast`, `write_back`, and `mode` options are passed to :func:`format_file_in_place` or :func:`format_stdin_to_stdout`. """ try: @@ -492,6 +496,36 @@ def reformat_one( report.failed(src, str(exc)) +def reformat_many( + sources: Set[Path], + fast: bool, + write_back: WriteBack, + mode: FileMode, + report: "Report", +) -> None: + """Reformat multiple files using a ProcessPoolExecutor.""" + loop = asyncio.get_event_loop() + worker_count = os.cpu_count() + if sys.platform == "win32": + # Work around https://bugs.python.org/issue26903 + worker_count = min(worker_count, 61) + executor = ProcessPoolExecutor(max_workers=worker_count) + try: + loop.run_until_complete( + schedule_formatting( + sources=sources, + fast=fast, + write_back=write_back, + mode=mode, + report=report, + loop=loop, + executor=executor, + ) + ) + finally: + shutdown(loop) + + async def schedule_formatting( sources: Set[Path], fast: bool, @@ -526,12 +560,14 @@ async def schedule_formatting( manager = Manager() lock = manager.Lock() tasks = { - loop.run_in_executor( - executor, format_file_in_place, src, fast, mode, write_back, lock + asyncio.ensure_future( + loop.run_in_executor( + executor, format_file_in_place, src, fast, mode, write_back, lock + ) ): src for src in sorted(sources) } - pending: Iterable[asyncio.Task] = tasks.keys() + pending: Iterable[asyncio.Future] = tasks.keys() try: loop.add_signal_handler(signal.SIGINT, cancel, pending) loop.add_signal_handler(signal.SIGTERM, cancel, pending) @@ -572,7 +608,7 @@ def format_file_in_place( 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`. + `mode` and `fast` options are passed to :func:`format_file_contents`. """ if src.suffix == ".pyi": mode = evolve(mode, is_pyi=True) @@ -650,7 +686,7 @@ def format_file_contents( If `fast` is False, additionally confirm that the reformatted code is valid by calling :func:`assert_equivalent` and :func:`assert_stable` on it. - `line_length` is passed to :func:`format_str`. + `mode` is passed to :func:`format_str`. """ if src_contents.strip() == "": raise NothingChanged @@ -668,10 +704,11 @@ def format_file_contents( def format_str(src_contents: str, *, mode: FileMode) -> FileContent: """Reformat a string and return new contents. - `line_length` determines how many characters per line are allowed. + `mode` determines formatting options, such as how many characters per line are + allowed. """ src_node = lib2to3_parse(src_contents.lstrip(), mode.target_versions) - dst_contents = "" + dst_contents = [] future_imports = get_future_imports(src_node) if mode.target_versions: versions = mode.target_versions @@ -694,15 +731,15 @@ def format_str(src_contents: str, *, mode: FileMode) -> FileContent: } for current_line in lines.visit(src_node): for _ in range(after): - dst_contents += str(empty_line) + dst_contents.append(str(empty_line)) before, after = elt.maybe_empty_lines(current_line) for _ in range(before): - dst_contents += str(empty_line) + dst_contents.append(str(empty_line)) for line in split_line( current_line, line_length=mode.line_length, features=split_line_features ): - dst_contents += str(line) - return dst_contents + dst_contents.append(str(line)) + return "".join(dst_contents) def decode_bytes(src: bytes) -> Tuple[FileContent, Encoding, NewLine]: @@ -726,16 +763,38 @@ def get_grammars(target_versions: Set[TargetVersion]) -> List[Grammar]: if not target_versions: # No target_version specified, so try all grammars. return [ + # Python 3.7+ + pygram.python_grammar_no_print_statement_no_exec_statement_async_keywords, + # Python 3.0-3.6 pygram.python_grammar_no_print_statement_no_exec_statement, + # Python 2.7 with future print_function import pygram.python_grammar_no_print_statement, + # Python 2.7 pygram.python_grammar, ] elif all(version.is_python2() for version in target_versions): # Python 2-only code, so try Python 2 grammars. - return [pygram.python_grammar_no_print_statement, pygram.python_grammar] + return [ + # Python 2.7 with future print_function import + pygram.python_grammar_no_print_statement, + # Python 2.7 + pygram.python_grammar, + ] else: # Python 3-compatible code, so only try Python 3 grammar. - return [pygram.python_grammar_no_print_statement_no_exec_statement] + grammars = [] + # If we have to parse both, try to parse async as a keyword first + if not supports_feature(target_versions, Feature.ASYNC_IDENTIFIERS): + # Python 3.7+ + grammars.append( + pygram.python_grammar_no_print_statement_no_exec_statement_async_keywords # noqa: B950 + ) + if not supports_feature(target_versions, Feature.ASYNC_KEYWORDS): + # Python 3.0-3.6 + grammars.append(pygram.python_grammar_no_print_statement_no_exec_statement) + # At least one of the above branches must have been taken, because every Python + # version has exactly one of the two 'ASYNC_*' flags + return grammars def lib2to3_parse(src_txt: str, target_versions: Iterable[TargetVersion] = ()) -> Node: @@ -1002,7 +1061,7 @@ class BracketTracker: """Return True if there is an yet unmatched open bracket on the line.""" return bool(self.bracket_match) - def max_delimiter_priority(self, exclude: Iterable[LeafID] = ()) -> int: + def max_delimiter_priority(self, exclude: Iterable[LeafID] = ()) -> Priority: """Return the highest priority of a delimiter found on the line. Values are consistent with what `is_split_*_delimiter()` return. @@ -1010,7 +1069,7 @@ class BracketTracker: """ return max(v for k, v in self.delimiters.items() if k not in exclude) - def delimiter_count_with_priority(self, priority: int = 0) -> int: + def delimiter_count_with_priority(self, priority: Priority = 0) -> int: """Return the number of delimiters with the given `priority`. If no `priority` is passed, defaults to max priority on the line. @@ -1223,10 +1282,13 @@ class Line: try: last_leaf = self.leaves[-1] ignored_ids.add(id(last_leaf)) - if last_leaf.type == token.COMMA: - # When trailing commas are inserted by Black for consistency, comments - # after the previous last element are not moved (they don't have to, - # rendering will still be correct). So we ignore trailing commas. + if last_leaf.type == token.COMMA or ( + last_leaf.type == token.RPAR and not last_leaf.value + ): + # When trailing commas or optional parens are inserted by Black for + # consistency, comments after the previous last element are not moved + # (they don't have to, rendering will still be correct). So we ignore + # trailing commas and invisible. last_leaf = self.leaves[-2] ignored_ids.add(id(last_leaf)) except IndexError: @@ -1293,7 +1355,10 @@ class Line: bracket_depth = leaf.bracket_depth if bracket_depth == depth and leaf.type == token.COMMA: commas += 1 - if leaf.parent and leaf.parent.type == syms.arglist: + if leaf.parent and leaf.parent.type in { + syms.arglist, + syms.typedargslist, + }: commas += 1 break @@ -1320,7 +1385,23 @@ class Line: comment.prefix = "" return False - self.comments.setdefault(id(self.leaves[-1]), []).append(comment) + last_leaf = self.leaves[-1] + if ( + last_leaf.type == token.RPAR + and not last_leaf.value + and last_leaf.parent + and len(list(last_leaf.parent.leaves())) <= 3 + and not is_type_comment(comment) + ): + # Comments on an optional parens wrapping a single leaf should belong to + # the wrapped node except if it's a type comment. Pinning the comment like + # this avoids unstable formatting caused by comment migration. + if len(self.leaves) < 2: + comment.type = STANDALONE_COMMENT + comment.prefix = "" + return False + last_leaf = self.leaves[-2] + self.comments.setdefault(id(last_leaf), []).append(comment) return True def comments_after(self, leaf: Leaf) -> List[Leaf]: @@ -1543,6 +1624,26 @@ class LineGenerator(Visitor[Line]): self.current_line.append(node) yield from super().visit_default(node) + def visit_atom(self, node: Node) -> Iterator[Line]: + # Always make parentheses invisible around a single node, because it should + # not be needed (except in the case of yield, where removing the parentheses + # produces a SyntaxError). + if ( + len(node.children) == 3 + and isinstance(node.children[0], Leaf) + and node.children[0].type == token.LPAR + and isinstance(node.children[2], Leaf) + and node.children[2].type == token.RPAR + and isinstance(node.children[1], Leaf) + and not ( + node.children[1].type == token.NAME + and node.children[1].value == "yield" + ) + ): + node.children[0].value = "" + node.children[2].value = "" + yield from super().visit_default(node) + def visit_INDENT(self, node: Node) -> Iterator[Line]: """Increase indentation level, maybe yield a line.""" # In blib2to3 INDENT never holds comments. @@ -1956,7 +2057,7 @@ def container_of(leaf: Leaf) -> LN: return container -def is_split_after_delimiter(leaf: Leaf, previous: Optional[Leaf] = None) -> int: +def is_split_after_delimiter(leaf: Leaf, previous: Optional[Leaf] = None) -> Priority: """Return the priority of the `leaf` delimiter, given a line break after it. The delimiter priorities returned here are from those delimiters that would @@ -1970,7 +2071,7 @@ def is_split_after_delimiter(leaf: Leaf, previous: Optional[Leaf] = None) -> int return 0 -def is_split_before_delimiter(leaf: Leaf, previous: Optional[Leaf] = None) -> int: +def is_split_before_delimiter(leaf: Leaf, previous: Optional[Leaf] = None) -> Priority: """Return the priority of the `leaf` delimiter, given a line break before it. The delimiter priorities returned here are from those delimiters that would @@ -2120,15 +2221,21 @@ def list_comments(prefix: str, *, is_endmarker: bool) -> List[ProtoComment]: consumed = 0 nlines = 0 + ignored_lines = 0 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: nlines += 1 if not line.startswith("#"): + # Escaped newlines outside of a comment are not really newlines at + # all. We treat a single-line comment following an escaped newline + # as a simple trailing comment. + if line.endswith("\\"): + ignored_lines += 1 continue - if index == 0 and not is_endmarker: + if index == ignored_lines and not is_endmarker: comment_type = token.COMMENT # simple trailing comment else: comment_type = STANDALONE_COMMENT @@ -2403,10 +2510,21 @@ def bracket_split_build_line( if leaves: # Since body is a new indent level, remove spurious leading whitespace. normalize_prefix(leaves[0], inside_brackets=True) - # Ensure a trailing comma when expected. - if original.is_import: - if leaves[-1].type != token.COMMA: - leaves.append(Leaf(token.COMMA, ",")) + # Ensure a trailing comma for imports and standalone function arguments, but + # be careful not to add one after any comments. + no_commas = original.is_def and not any( + l.type == token.COMMA for l in leaves + ) + + if original.is_import or no_commas: + for i in range(len(leaves) - 1, -1, -1): + if leaves[i].type == STANDALONE_COMMENT: + continue + elif leaves[i].type == token.COMMA: + break + else: + leaves.insert(i + 1, Leaf(token.COMMA, ",")) + break # Populate the line for leaf in leaves: result.append(leaf, preformatted=True) @@ -2637,7 +2755,15 @@ def normalize_string_quotes(leaf: Leaf) -> None: 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 "f" in prefix.casefold(): - matches = re.findall(r"[^{]\{(.*?)\}[^}]", new_body) + matches = re.findall( + r""" + (?:[^{]|^)\{ # start of the string or a non-{ followed by a single { + ([^{].*?) # contents of the brackets except if begins with {{ + \}(?:[^}]|$) # A } followed by end of the string or a non-} + """, + new_body, + re.VERBOSE, + ) for m in matches: if "\\" in str(m): # Do not introduce backslashes in interpolated expressions @@ -2704,7 +2830,7 @@ def format_float_or_int_string(text: str) -> str: def normalize_invisible_parens(node: Node, parens_after: Set[str]) -> None: """Make existing optional parentheses invisible or create new ones. - `parens_after` is a set of string leaf values immeditely after which parens + `parens_after` is a set of string leaf values immediately after which parens should be put. Standardizes on visible parentheses for single-element tuples, and keeps @@ -2717,9 +2843,17 @@ def normalize_invisible_parens(node: Node, parens_after: Set[str]) -> None: check_lpar = False for index, child in enumerate(list(node.children)): + # Add parentheses around long tuple unpacking in assignments. + if ( + index == 0 + and isinstance(child, Node) + and child.type == syms.testlist_star_expr + ): + check_lpar = True + if check_lpar: if child.type == syms.atom: - if maybe_make_parens_invisible_in_atom(child): + if maybe_make_parens_invisible_in_atom(child, parent=node): lpar = Leaf(token.LPAR, "") rpar = Leaf(token.RPAR, "") index = child.remove() or 0 @@ -2748,7 +2882,11 @@ def normalize_invisible_parens(node: Node, parens_after: Set[str]) -> None: lpar = Leaf(token.LPAR, "") rpar = Leaf(token.RPAR, "") index = child.remove() or 0 - node.insert_child(index, Node(syms.atom, [lpar, child, rpar])) + prefix = child.prefix + child.prefix = "" + new_child = Node(syms.atom, [lpar, child, rpar]) + new_child.prefix = prefix + node.insert_child(index, new_child) check_lpar = isinstance(child, Leaf) and child.value in parens_after @@ -2830,7 +2968,7 @@ def generate_ignored_nodes(leaf: Leaf) -> Iterator[LN]: container = container.next_sibling -def maybe_make_parens_invisible_in_atom(node: LN) -> bool: +def maybe_make_parens_invisible_in_atom(node: LN, parent: LN) -> bool: """If it's safe, make the parens in the atom `node` invisible, recursively. Returns whether the node should itself be wrapped in invisible parentheses. @@ -2840,7 +2978,7 @@ def maybe_make_parens_invisible_in_atom(node: LN) -> bool: node.type != syms.atom or is_empty_tuple(node) or is_one_tuple(node) - or is_yield(node) + or (is_yield(node) and parent.type != syms.expr_stmt) or max_delimiter_priority_in_atom(node) >= COMMA_PRIORITY ): return False @@ -2852,7 +2990,7 @@ def maybe_make_parens_invisible_in_atom(node: LN) -> bool: first.value = "" # type: ignore last.value = "" # type: ignore if len(node.children) > 1: - maybe_make_parens_invisible_in_atom(node.children[1]) + maybe_make_parens_invisible_in_atom(node.children[1], parent=parent) return False return True @@ -2970,7 +3108,7 @@ def is_stub_body(node: LN) -> bool: ) -def max_delimiter_priority_in_atom(node: LN) -> int: +def max_delimiter_priority_in_atom(node: LN) -> Priority: """Return maximum delimiter priority inside `node`. This is specific to atoms with contents contained in a pair of parentheses. @@ -3330,17 +3468,32 @@ class Report: return ", ".join(report) + "." +def parse_ast(src: str) -> Union[ast3.AST, ast27.AST]: + for feature_version in (7, 6): + try: + return ast3.parse(src, feature_version=feature_version) + except SyntaxError: + continue + + return ast27.parse(src) + + def assert_equivalent(src: str, dst: str) -> None: """Raise AssertionError if `src` and `dst` aren't equivalent.""" - import ast - import traceback - - def _v(node: ast.AST, depth: int = 0) -> Iterator[str]: + def _v(node: Union[ast3.AST, ast27.AST], depth: int = 0) -> Iterator[str]: """Simple visitor generating strings to compare ASTs by content.""" yield f"{' ' * depth}{node.__class__.__name__}(" for field in sorted(node._fields): + # TypeIgnore has only one field 'lineno' which breaks this comparison + if isinstance(node, (ast3.TypeIgnore, ast27.TypeIgnore)): + break + + # Ignore str kind which is case sensitive / and ignores unicode_literals + if isinstance(node, (ast3.Str, ast27.Str, ast3.Bytes)) and field == "kind": + continue + try: value = getattr(node, field) except AttributeError: @@ -3354,15 +3507,15 @@ def assert_equivalent(src: str, dst: str) -> None: # parentheses and they change the AST. if ( field == "targets" - and isinstance(node, ast.Delete) - and isinstance(item, ast.Tuple) + and isinstance(node, (ast3.Delete, ast27.Delete)) + and isinstance(item, (ast3.Tuple, ast27.Tuple)) ): for item in item.elts: yield from _v(item, depth + 2) - elif isinstance(item, ast.AST): + elif isinstance(item, (ast3.AST, ast27.AST)): yield from _v(item, depth + 2) - elif isinstance(value, ast.AST): + elif isinstance(value, (ast3.AST, ast27.AST)): yield from _v(value, depth + 2) else: @@ -3371,17 +3524,15 @@ def assert_equivalent(src: str, dst: str) -> None: yield f"{' ' * depth}) # /{node.__class__.__name__}" try: - src_ast = ast.parse(src) + src_ast = parse_ast(src) except Exception as exc: - major, minor = sys.version_info[:2] raise AssertionError( - f"cannot use --safe with this file; failed to parse source file " - f"with Python {major}.{minor}'s builtin AST. Re-run with --fast " - f"or stop using deprecated Python 2 syntax. AST error message: {exc}" + f"cannot use --safe with this file; failed to parse source file. " + f"AST error message: {exc}" ) try: - dst_ast = ast.parse(dst) + dst_ast = parse_ast(dst) except Exception as exc: log = dump_to_file("".join(traceback.format_tb(exc.__traceback__)), dst) raise AssertionError( @@ -3420,8 +3571,6 @@ def assert_stable(src: str, dst: str, mode: FileMode) -> None: def dump_to_file(*output: str) -> str: """Dump `output` to a temporary file. Return path to the file.""" - import tempfile - with tempfile.NamedTemporaryFile( mode="w", prefix="blk_", suffix=".log", delete=False, encoding="utf8" ) as f: @@ -3519,7 +3668,6 @@ def enumerate_with_length( if "\n" in leaf.value: return # Multiline strings, we can't continue. - comment: Optional[Leaf] for comment in line.comments_after(leaf): length += len(comment.value)