X-Git-Url: https://git.madduck.net/etc/vim.git/blobdiff_plain/ee02ebe3e9996345acd4c042f7b8daffb686167b..affc0ec3f90803e77fbba62376e7be9b71fe0807:/black.py?ds=inline diff --git a/black.py b/black.py index 19a023c..e72db2b 100644 --- a/black.py +++ b/black.py @@ -1,7 +1,7 @@ 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 import io @@ -10,10 +10,11 @@ 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, @@ -45,7 +46,7 @@ from blib2to3.pgen2 import driver, token from blib2to3.pgen2.parse import ParseError -__version__ = "18.5b1" +__version__ = "18.6b1" DEFAULT_LINE_LENGTH = 88 DEFAULT_EXCLUDES = ( r"/(\.git|\.hg|\.mypy_cache|\.tox|\.venv|_build|buck-out|build|dist)/" @@ -57,6 +58,7 @@ CACHE_DIR = Path(user_cache_dir("black", version=__version__)) # types FileContent = str Encoding = str +NewLine = str Depth = int NodeType = int LeafID = int @@ -335,8 +337,9 @@ def main( finally: shutdown(loop) if verbose or not quiet: - out("All done! ✨ 🍰 ✨") - click.echo(str(report)) + bang = "💥 💔 💥" if report.return_code else "✨ 🍰 ✨" + out(f"All done! {bang}") + click.secho(str(report), err=True) ctx.exit(report.return_code) @@ -468,8 +471,9 @@ def format_file_in_place( if src.suffix == ".pyi": mode |= FileMode.PYI + then = datetime.utcfromtimestamp(src.stat().st_mtime) with open(src, "rb") as buf: - newline, encoding, src_contents = prepare_input(buf.read()) + src_contents, encoding, newline = decode_bytes(buf.read()) try: dst_contents = format_file_contents( src_contents, line_length=line_length, fast=fast, mode=mode @@ -481,8 +485,9 @@ def format_file_in_place( 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() @@ -513,7 +518,8 @@ def format_stdin_to_stdout( `line_length`, `fast`, `is_pyi`, and `force_py36` arguments are passed to :func:`format_file_contents`. """ - newline, encoding, src = prepare_input(sys.stdin.buffer.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) @@ -523,26 +529,17 @@ 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: - f = io.TextIOWrapper( - sys.stdout.buffer, - encoding=encoding, - newline=newline, - write_through=True, - ) f.write(dst) - f.detach() elif write_back == WriteBack.DIFF: - src_name = " (original)" - dst_name = " (formatted)" - f = io.TextIOWrapper( - sys.stdout.buffer, - encoding=encoding, - newline=newline, - write_through=True, - ) + 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() + f.detach() def format_file_contents( @@ -603,17 +600,21 @@ def format_str( return dst_contents -def prepare_input(src: bytes) -> Tuple[str, str, str]: - """Analyze `src` and return a tuple of (newline, encoding, decoded_contents) +def decode_bytes(src: bytes) -> Tuple[FileContent, Encoding, NewLine]: + """Return a tuple of (decoded_contents, encoding, newline). - Where `newline` is either CRLF or LF, and `decoded_contents` is decoded with - universal newlines (i.e. only LF). + `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) - return newline, encoding, io.TextIOWrapper(srcbuf, encoding).read() + with io.TextIOWrapper(srcbuf, encoding) as tiow: + return tiow.read(), encoding, newline GRAMMARS = [ @@ -626,7 +627,7 @@ 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": + if src_txt[-1:] != "\n": src_txt += "\n" for grammar in GRAMMARS: drv = driver.Driver(grammar, pytree.convert) @@ -771,6 +772,7 @@ UNPACKING_PARENTS = { syms.dictsetmaker, syms.listmaker, syms.testlist_gexp, + syms.testlist_star_expr, } TEST_DESCENDANTS = { syms.test, @@ -1093,6 +1095,13 @@ 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 ( @@ -1826,7 +1835,7 @@ 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: @@ -2212,32 +2221,50 @@ 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 @@ -2535,7 +2562,7 @@ def normalize_invisible_parens(node: Node, parens_after: Set[str]) -> None: 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) @@ -3182,6 +3209,42 @@ 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?