X-Git-Url: https://git.madduck.net/etc/vim.git/blobdiff_plain/90e14e8b619246ff787d10b94bd5a64d18e52b84..6695fd892f8f1a9e0d3879423bb0361e07eedd24:/black.py diff --git a/black.py b/black.py index b2cf543..2668747 100644 --- a/black.py +++ b/black.py @@ -1,60 +1,89 @@ +import ast import asyncio -import pickle -from asyncio.base_events import BaseEventLoop -from concurrent.futures import Executor, ProcessPoolExecutor +from abc import ABC, abstractmethod +from collections import defaultdict +from concurrent.futures import Executor, ThreadPoolExecutor, ProcessPoolExecutor +from contextlib import contextmanager +from datetime import datetime from enum import Enum -from functools import partial, wraps -import keyword +from functools import lru_cache, partial, wraps +import io +import itertools import logging -from multiprocessing import Manager +from multiprocessing import Manager, freeze_support import os from pathlib import Path -import re -import tokenize +import pickle +import regex as re import signal import sys +import tempfile +import tokenize +import traceback from typing import ( Any, Callable, Collection, Dict, + Generator, Generic, Iterable, Iterator, List, Optional, Pattern, + Sequence, Set, Tuple, Type, TypeVar, Union, + cast, + TYPE_CHECKING, ) +from typing_extensions import Final +from mypy_extensions import mypyc_attr from appdirs import user_cache_dir -from attr import dataclass, Factory +from dataclasses import dataclass, field, replace import click +import toml +from typed_ast import ast3, ast27 +from pathspec import PathSpec # lib2to3 fork from blib2to3.pytree import Node, Leaf, type_repr from blib2to3 import pygram, pytree from blib2to3.pgen2 import driver, token +from blib2to3.pgen2.grammar import Grammar from blib2to3.pgen2.parse import ParseError -__version__ = "18.4a6" +from _black_version import version as __version__ + +if TYPE_CHECKING: + import colorama # noqa: F401 + DEFAULT_LINE_LENGTH = 88 +DEFAULT_EXCLUDES = r"/(\.eggs|\.git|\.hg|\.mypy_cache|\.nox|\.tox|\.venv|\.svn|_build|buck-out|build|dist)/" # noqa: B950 +DEFAULT_INCLUDES = r"\.pyi?$" +CACHE_DIR = Path(user_cache_dir("black", version=__version__)) + +STRING_PREFIX_CHARS: Final = "furbFURB" # All possible string prefix characters. + # types -syms = pygram.python_symbols FileContent = str Encoding = str +NewLine = str Depth = int NodeType = int +ParserState = int LeafID = int +StringID = int Priority = int Index = int LN = Union[Leaf, Node] -SplitFunc = Callable[["Line", bool], Iterator["Line"]] +Transformer = Callable[["Line", Collection["Feature"]], Iterator["Line"]] Timestamp = float FileSize = int CacheInfo = Tuple[Timestamp, FileSize] @@ -62,51 +91,72 @@ Cache = Dict[Path, CacheInfo] out = partial(click.secho, bold=True, err=True) err = partial(click.secho, fg="red", err=True) +pygram.initialize(CACHE_DIR) +syms = pygram.python_symbols + class NothingChanged(UserWarning): - """Raised by :func:`format_file` when reformatted code is the same as source.""" + """Raised when reformatted code is the same as source.""" -class CannotSplit(Exception): - """A readable split that fits the allotted line length is impossible. +class CannotTransform(Exception): + """Base class for errors raised by Transformers.""" - Raised by :func:`left_hand_split`, :func:`right_hand_split`, and - :func:`delimiter_split`. - """ +class CannotSplit(CannotTransform): + """A readable split that fits the allotted line length is impossible.""" -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. - """ +class InvalidInput(ValueError): + """Raised when input source code fails all parse attempts.""" + + +T = TypeVar("T") +E = TypeVar("E", bound=Exception) + - def __init__(self, consumed: int) -> None: - super().__init__(consumed) - self.consumed = consumed +class Ok(Generic[T]): + def __init__(self, value: T) -> None: + self._value = value - def trim_prefix(self, leaf: Leaf) -> None: - leaf.prefix = leaf.prefix[self.consumed :] + def ok(self) -> T: + return self._value - 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 Err(Generic[E]): + def __init__(self, e: E) -> None: + self._e = e -class FormatOn(FormatError): - """Found a comment like `# fmt: on` in the file.""" + def err(self) -> E: + return self._e -class FormatOff(FormatError): - """Found a comment like `# fmt: off` in the file.""" +# The 'Result' return type is used to implement an error-handling model heavily +# influenced by that used by the Rust programming language +# (see https://doc.rust-lang.org/book/ch09-00-error-handling.html). +Result = Union[Ok[T], Err[E]] +TResult = Result[T, CannotTransform] # (T)ransform Result +TMatchResult = TResult[Index] class WriteBack(Enum): NO = 0 YES = 1 DIFF = 2 + CHECK = 3 + COLOR_DIFF = 4 + + @classmethod + def from_configuration( + cls, *, check: bool, diff: bool, color: bool = False + ) -> "WriteBack": + if check and not diff: + return cls.CHECK + + if diff and color: + return cls.COLOR_DIFF + + return cls.DIFF if diff else cls.YES class Changed(Enum): @@ -115,22 +165,224 @@ class Changed(Enum): YES = 2 -@click.command() +class TargetVersion(Enum): + PY27 = 2 + PY33 = 3 + PY34 = 4 + PY35 = 5 + PY36 = 6 + PY37 = 7 + PY38 = 8 + + def is_python2(self) -> bool: + return self is TargetVersion.PY27 + + +PY36_VERSIONS = {TargetVersion.PY36, TargetVersion.PY37, TargetVersion.PY38} + + +class Feature(Enum): + # All string literals are unicode + UNICODE_LITERALS = 1 + F_STRINGS = 2 + 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 + ASSIGNMENT_EXPRESSIONS = 8 + POS_ONLY_ARGUMENTS = 9 + + +VERSION_TO_FEATURES: Dict[TargetVersion, Set[Feature]] = { + 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, + Feature.F_STRINGS, + Feature.NUMERIC_UNDERSCORES, + Feature.TRAILING_COMMA_IN_CALL, + Feature.TRAILING_COMMA_IN_DEF, + Feature.ASYNC_KEYWORDS, + }, + TargetVersion.PY38: { + Feature.UNICODE_LITERALS, + Feature.F_STRINGS, + Feature.NUMERIC_UNDERSCORES, + Feature.TRAILING_COMMA_IN_CALL, + Feature.TRAILING_COMMA_IN_DEF, + Feature.ASYNC_KEYWORDS, + Feature.ASSIGNMENT_EXPRESSIONS, + Feature.POS_ONLY_ARGUMENTS, + }, +} + + +@dataclass +class Mode: + target_versions: Set[TargetVersion] = field(default_factory=set) + line_length: int = DEFAULT_LINE_LENGTH + string_normalization: bool = True + is_pyi: bool = False + + def get_cache_key(self) -> str: + if self.target_versions: + version_str = ",".join( + str(version.value) + for version in sorted(self.target_versions, key=lambda v: v.value) + ) + else: + version_str = "-" + parts = [ + version_str, + str(self.line_length), + str(int(self.string_normalization)), + str(int(self.is_pyi)), + ] + return ".".join(parts) + + +# Legacy name, left for integrations. +FileMode = Mode + + +def supports_feature(target_versions: Set[TargetVersion], feature: Feature) -> bool: + return all(feature in VERSION_TO_FEATURES[version] for version in target_versions) + + +def find_pyproject_toml(path_search_start: str) -> Optional[str]: + """Find the absolute filepath to a pyproject.toml if it exists""" + path_project_root = find_project_root(path_search_start) + path_pyproject_toml = path_project_root / "pyproject.toml" + return str(path_pyproject_toml) if path_pyproject_toml.is_file() else None + + +def parse_pyproject_toml(path_config: str) -> Dict[str, Any]: + """Parse a pyproject toml file, pulling out relevant parts for Black + + If parsing fails, will raise a toml.TomlDecodeError + """ + pyproject_toml = toml.load(path_config) + config = pyproject_toml.get("tool", {}).get("black", {}) + return {k.replace("--", "").replace("-", "_"): v for k, v in config.items()} + + +def read_pyproject_toml( + ctx: click.Context, param: click.Parameter, value: Optional[str] +) -> 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. + """ + if not value: + value = find_pyproject_toml(ctx.params.get("src", ())) + if value is None: + return None + + try: + config = parse_pyproject_toml(value) + except (toml.TomlDecodeError, OSError) as e: + raise click.FileError( + filename=value, hint=f"Error reading configuration file: {e}" + ) + + if not config: + return None + + target_version = config.get("target_version") + if target_version is not None and not isinstance(target_version, list): + raise click.BadOptionUsage( + "target-version", f"Config key target-version must be a list" + ) + + default_map: Dict[str, Any] = {} + if ctx.default_map: + default_map.update(ctx.default_map) + default_map.update(config) + + ctx.default_map = default_map + return value + + +def target_version_option_callback( + c: click.Context, p: Union[click.Option, click.Parameter], v: Tuple[str, ...] +) -> List[TargetVersion]: + """Compute the target versions from a --target-version flag. + + This is its own function because mypy couldn't infer the type correctly + when it was a lambda, causing mypyc trouble. + """ + return [TargetVersion[val.upper()] for val in v] + + +@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", 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( + "-t", + "--target-version", + type=click.Choice([v.name.lower() for v in TargetVersion]), + callback=target_version_option_callback, + multiple=True, + help=( + "Python versions that should be supported by Black's output. [default: per-file" + " auto-detection]" + ), +) +@click.option( + "--py36", + is_flag=True, + help=( + "Allow using Python 3.6-only syntax on all input files. This will put trailing" + " commas in function signatures and calls also after *args and **kwargs." + " Deprecated; use --target-version instead. [default: per-file auto-detection]" + ), +) +@click.option( + "--pyi", + is_flag=True, + help=( + "Format all input files like typing stubs regardless of file extension (useful" + " when piping source on standard input)." + ), +) +@click.option( + "-S", + "--skip-string-normalization", + is_flag=True, + help="Don't normalize string quotes or prefixes.", +) @click.option( "--check", is_flag=True, help=( - "Don't write the files back, just return the status. Return code 0 " - "means nothing would change. Return code 1 means some files would be " - "reformatted. Return code 123 means there was an internal error." + "Don't write the files back, just return the status. Return code 0 means" + " nothing would change. Return code 1 means some files would be reformatted." + " Return code 123 means there was an internal error." ), ) @click.option( @@ -138,18 +390,56 @@ class Changed(Enum): is_flag=True, help="Don't write the files back, just output a diff for each file on stdout.", ) +@click.option( + "--color/--no-color", + is_flag=True, + help="Show colored diff. Only applies when `--diff` is given.", +) @click.option( "--fast/--safe", is_flag=True, help="If --fast given, skip temporary sanity checks. [default: --safe]", ) +@click.option( + "--include", + type=str, + default=DEFAULT_INCLUDES, + help=( + "A regular expression that matches files and directories that should be" + " included on recursive searches. An empty value means all files are included" + " regardless of the name. Use forward slashes for directories on all platforms" + " (Windows, too). Exclusions are calculated first, inclusions later." + ), + show_default=True, +) +@click.option( + "--exclude", + type=str, + default=DEFAULT_EXCLUDES, + help=( + "A regular expression that matches files and directories that should be" + " excluded on recursive searches. An empty value means no paths are excluded." + " Use forward slashes for directories on all platforms (Windows, too). " + " Exclusions are calculated first, inclusions later." + ), + show_default=True, +) @click.option( "-q", "--quiet", is_flag=True, help=( - "Don't emit non-error messages to stderr. Errors are still emitted, " - "silence those with 2>/dev/null." + "Don't emit non-error messages to stderr. Errors are still emitted; 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__) @@ -159,286 +449,552 @@ class Changed(Enum): 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=True, + file_okay=True, + dir_okay=False, + readable=True, + allow_dash=False, + path_type=str, + ), + is_eager=True, + callback=read_pyproject_toml, + help="Read configuration from PATH.", ) @click.pass_context def main( ctx: click.Context, + code: Optional[str], line_length: int, + target_version: List[TargetVersion], check: bool, diff: bool, + color: bool, fast: bool, + pyi: bool, + py36: bool, + skip_string_normalization: bool, quiet: bool, - src: List[str], + verbose: bool, + include: str, + exclude: str, + src: Tuple[str, ...], + config: Optional[str], ) -> None: """The uncompromising code formatter.""" - sources: List[Path] = [] + write_back = WriteBack.from_configuration(check=check, diff=diff, color=color) + if target_version: + if py36: + err("Cannot use both --target-version and --py36") + ctx.exit(2) + else: + versions = set(target_version) + elif py36: + err( + "--py36 is deprecated and will be removed in a future version. Use" + " --target-version py36 instead." + ) + versions = PY36_VERSIONS + else: + # We'll autodetect later. + versions = set() + mode = Mode( + target_versions=versions, + line_length=line_length, + is_pyi=pyi, + string_normalization=not skip_string_normalization, + ) + 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: + err(f"Invalid regular expression for include given: {include!r}") + ctx.exit(2) + try: + 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, diff=diff, quiet=quiet, verbose=verbose) + root = find_project_root(src) + sources: Set[Path] = set() + path_empty(src, quiet, verbose, ctx) for s in src: p = Path(s) if p.is_dir(): - sources.extend(gen_python_files_in_dir(p)) - elif p.is_file(): + sources.update( + gen_python_files_in_dir( + p, root, include_regex, exclude_regex, report, get_gitignore(root) + ) + ) + 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 - report = Report(check=check, quiet=quiet) if len(sources) == 0: - out("No paths given. Nothing to do 😴") + if verbose or not quiet: + out("No Python files are present to be formatted. Nothing to do 😴") ctx.exit(0) - return - elif len(sources) == 1: - reformat_one(sources[0], line_length, fast, write_back, report) + if len(sources) == 1: + reformat_one( + src=sources.pop(), + fast=fast, + write_back=write_back, + mode=mode, + report=report, + ) else: - loop = asyncio.get_event_loop() - executor = ProcessPoolExecutor(max_workers=os.cpu_count()) - try: - loop.run_until_complete( - schedule_formatting( - sources, line_length, fast, write_back, report, loop, executor - ) - ) - finally: - shutdown(loop) - if not quiet: - out("All done! ✨ 🍰 ✨") - click.echo(str(report)) + reformat_many( + sources=sources, fast=fast, write_back=write_back, mode=mode, report=report + ) + + if verbose or not quiet: + out("Oh no! 💥 💔 💥" if report.return_code else "All done! ✨ 🍰 ✨") + click.secho(str(report), err=True) ctx.exit(report.return_code) +def path_empty( + src: Tuple[str, ...], quiet: bool, verbose: bool, ctx: click.Context +) -> None: + """ + Exit if there is no `src` provided for formatting + """ + if not src: + if verbose or not quiet: + out("No Path provided. Nothing to do 😴") + ctx.exit(0) + + def reformat_one( - src: Path, line_length: int, fast: bool, write_back: WriteBack, report: "Report" + src: Path, fast: bool, write_back: WriteBack, mode: Mode, report: "Report" ) -> 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`, and `fast` options are passed to :func:`format_file_in_place`. + `fast`, `write_back`, and `mode` options are passed to + :func:`format_file_in_place` or :func:`format_stdin_to_stdout`. """ 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 - ): + if format_stdin_to_stdout(fast=fast, write_back=write_back, mode=mode): changed = Changed.YES else: cache: Cache = {} if write_back != WriteBack.DIFF: - cache = read_cache(line_length) - src = src.resolve() - if src in cache and cache[src] == get_cache_info(src): + cache = read_cache(mode) + res_src = src.resolve() + if res_src in cache and cache[res_src] == get_cache_info(res_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 - ) + if changed is not Changed.CACHED and format_file_in_place( + src, fast=fast, write_back=write_back, mode=mode ): changed = Changed.YES - if write_back == WriteBack.YES and changed is not Changed.NO: - write_cache(cache, [src], line_length) + 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], mode) report.done(src, changed) except Exception as exc: report.failed(src, str(exc)) +def reformat_many( + sources: Set[Path], fast: bool, write_back: WriteBack, mode: Mode, report: "Report" +) -> None: + """Reformat multiple files using a ProcessPoolExecutor.""" + executor: Executor + 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) + try: + executor = ProcessPoolExecutor(max_workers=worker_count) + except OSError: + # we arrive here if the underlying system does not support multi-processing + # like in AWS Lambda, in which case we gracefully fallback to + # a ThreadPollExecutor with just a single worker (more workers would not do us + # any good due to the Global Interpreter Lock) + executor = ThreadPoolExecutor(max_workers=1) + + 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) + if executor is not None: + executor.shutdown() + + async def schedule_formatting( - sources: List[Path], - line_length: int, + sources: Set[Path], fast: bool, write_back: WriteBack, + mode: Mode, report: "Report", - loop: BaseEventLoop, + loop: asyncio.AbstractEventLoop, executor: Executor, ) -> None: """Run formatting of `sources` in parallel using the provided `executor`. (Use ProcessPoolExecutors for actual parallelism.) - `line_length`, `write_back`, and `fast` options are passed to + `write_back`, `fast`, and `mode` options are passed to :func:`format_file_in_place`. """ cache: Cache = {} if write_back != WriteBack.DIFF: - cache = read_cache(line_length) + cache = read_cache(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 = { - src: loop.run_in_executor( - executor, format_file_in_place, src, line_length, fast, write_back, lock + 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 = { + asyncio.ensure_future( + loop.run_in_executor( + executor, format_file_in_place, src, fast, mode, 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(): + ): src + for src in sorted(sources) + } + pending: Iterable["asyncio.Future[bool]"] = 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) - + 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) + if sources_to_cache: + write_cache(cache, sources_to_cache, mode) def format_file_in_place( src: Path, - line_length: int, fast: bool, + mode: Mode, write_back: WriteBack = WriteBack.NO, lock: Any = None, # multiprocessing.Manager().Lock() is some crazy proxy ) -> bool: """Format file under `src` path. Return True if changed. - If `write_back` is True, write reformatted code back to stdout. - `line_length` and `fast` options are passed to :func:`format_file_contents`. + If `write_back` is DIFF, write a diff to stdout. If it is YES, write reformatted + code to the file. + `mode` and `fast` options are passed to :func:`format_file_contents`. """ + if src.suffix == ".pyi": + mode = replace(mode, is_pyi=True) - 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 - ) + dst_contents = format_file_contents(src_contents, fast=fast, mode=mode) except NothingChanged: return False - if write_back == write_back.YES: - with open(src, "w", encoding=src_buffer.encoding) as f: + if write_back == WriteBack.YES: + 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)" + elif write_back in (WriteBack.DIFF, WriteBack.COLOR_DIFF): + 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) - finally: - if lock: - lock.release() + + if write_back == write_back.COLOR_DIFF: + diff_contents = color_diff(diff_contents) + + with lock or nullcontext(): + f = io.TextIOWrapper( + sys.stdout.buffer, + encoding=encoding, + newline=newline, + write_through=True, + ) + f = wrap_stream_for_windows(f) + f.write(diff_contents) + f.detach() + return True +def color_diff(contents: str) -> str: + """Inject the ANSI color codes to the diff.""" + lines = contents.split("\n") + for i, line in enumerate(lines): + if line.startswith("+++") or line.startswith("---"): + line = "\033[1;37m" + line + "\033[0m" # bold white, reset + if line.startswith("@@"): + line = "\033[36m" + line + "\033[0m" # cyan, reset + if line.startswith("+"): + line = "\033[32m" + line + "\033[0m" # green, reset + elif line.startswith("-"): + line = "\033[31m" + line + "\033[0m" # red, reset + lines[i] = line + return "\n".join(lines) + + +def wrap_stream_for_windows( + f: io.TextIOWrapper, +) -> Union[io.TextIOWrapper, "colorama.AnsiToWin32.AnsiToWin32"]: + """ + Wrap the stream in colorama's wrap_stream so colors are shown on Windows. + + If `colorama` is not found, then no change is made. If `colorama` does + exist, then it handles the logic to determine whether or not to change + things. + """ + try: + from colorama import initialise + + # We set `strip=False` so that we can don't have to modify + # test_express_diff_with_color. + f = initialise.wrap_stream( + f, convert=None, strip=False, autoreset=False, wrap=True + ) + + # wrap_stream returns a `colorama.AnsiToWin32.AnsiToWin32` object + # which does not have a `detach()` method. So we fake one. + f.detach = lambda *args, **kwargs: None # type: ignore + except ImportError: + pass + + return f + + def format_stdin_to_stdout( - line_length: int, fast: bool, write_back: WriteBack = WriteBack.NO + fast: bool, *, write_back: WriteBack = WriteBack.NO, mode: Mode ) -> bool: """Format file on stdin. Return True if changed. - If `write_back` is True, write reformatted code back to stdout. - `line_length` and `fast` arguments are passed to :func:`format_file_contents`. + If `write_back` is YES, write reformatted code back to stdout. If it is DIFF, + write a diff to stdout. The `mode` argument is 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) + dst = format_file_contents(src, fast=fast, mode=mode) return True except NothingChanged: 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) - elif write_back == WriteBack.DIFF: - src_name = " (original)" - dst_name = " (formatted)" - sys.stdout.write(diff(src, dst, src_name, dst_name)) - - -def format_file_contents( - src_contents: str, line_length: int, fast: bool -) -> FileContent: + f.write(dst) + elif write_back in (WriteBack.DIFF, WriteBack.COLOR_DIFF): + now = datetime.utcnow() + src_name = f"STDIN\t{then} +0000" + dst_name = f"STDOUT\t{now} +0000" + d = diff(src, dst, src_name, dst_name) + if write_back == WriteBack.COLOR_DIFF: + d = color_diff(d) + f = wrap_stream_for_windows(f) + f.write(d) + f.detach() + + +def format_file_contents(src_contents: str, *, fast: bool, mode: Mode) -> FileContent: """Reformat contents a file and return new 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 - dst_contents = format_str(src_contents, line_length=line_length) + dst_contents = format_str(src_contents, mode=mode) if src_contents == dst_contents: raise NothingChanged if not fast: assert_equivalent(src_contents, dst_contents) - assert_stable(src_contents, dst_contents, line_length=line_length) + assert_stable(src_contents, dst_contents, mode=mode) return dst_contents -def format_str(src_contents: str, line_length: int) -> FileContent: +def format_str(src_contents: str, *, mode: Mode) -> 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. Example: + + >>> import black + >>> print(black.format_str("def f(arg:str='')->None:...", mode=Mode())) + def f(arg: str = "") -> None: + ... + + A more complex example: + >>> print( + ... black.format_str( + ... "def f(arg:str='')->None: hey", + ... mode=black.Mode( + ... target_versions={black.TargetVersion.PY36}, + ... line_length=10, + ... string_normalization=False, + ... is_pyi=False, + ... ), + ... ), + ... ) + def f( + arg: str = '', + ) -> None: + hey + """ - src_node = lib2to3_parse(src_contents) - dst_contents = "" - lines = LineGenerator() - elt = EmptyLineTracker() - py36 = is_python36(src_node) + src_node = lib2to3_parse(src_contents.lstrip(), mode.target_versions) + dst_contents = [] + future_imports = get_future_imports(src_node) + if mode.target_versions: + versions = mode.target_versions + else: + versions = detect_target_versions(src_node) + normalize_fmt_off(src_node) + lines = LineGenerator( + remove_u_prefix="unicode_literals" in future_imports + or supports_feature(versions, Feature.UNICODE_LITERALS), + is_pyi=mode.is_pyi, + normalize_strings=mode.string_normalization, + ) + elt = EmptyLineTracker(is_pyi=mode.is_pyi) empty_line = Line() after = 0 + split_line_features = { + feature + for feature in {Feature.TRAILING_COMMA_IN_CALL, Feature.TRAILING_COMMA_IN_DEF} + if supports_feature(versions, feature) + } for current_line in lines.visit(src_node): - for _ in range(after): - dst_contents += str(empty_line) + dst_contents.append(str(empty_line) * after) before, after = elt.maybe_empty_lines(current_line) - for _ in range(before): - dst_contents += str(empty_line) - for line in split_line(current_line, line_length=line_length, py36=py36): - dst_contents += str(line) - return dst_contents + dst_contents.append(str(empty_line) * before) + for line in transform_line( + current_line, + line_length=mode.line_length, + normalize_strings=mode.string_normalization, + features=split_line_features, + ): + dst_contents.append(str(line)) + return "".join(dst_contents) + +def decode_bytes(src: bytes) -> Tuple[FileContent, Encoding, NewLine]: + """Return a tuple of (decoded_contents, encoding, newline). -GRAMMARS = [ - pygram.python_grammar_no_print_statement_no_exec_statement, - pygram.python_grammar_no_print_statement, - pygram.python_grammar, -] + `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 + + +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, + ] + + if all(version.is_python2() for version in target_versions): + # Python 2-only code, so try Python 2 grammars. + return [ + # Python 2.7 with future print_function import + pygram.python_grammar_no_print_statement, + # Python 2.7 + pygram.python_grammar, + ] + + # Python 3-compatible code, so only try Python 3 grammar. + 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 + ) + 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) -> Node: +def lib2to3_parse(src_txt: str, target_versions: Iterable[TargetVersion] = ()) -> 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 - for grammar in GRAMMARS: + if src_txt[-1:] != "\n": + src_txt += "\n" + + for grammar in get_grammars(set(target_versions)): drv = driver.Driver(grammar, pytree.convert) try: result = drv.parse_string(src_txt, True) @@ -451,7 +1007,7 @@ def lib2to3_parse(src_txt: str) -> Node: faulty_line = lines[lineno - 1] except IndexError: faulty_line = "" - exc = ValueError(f"Cannot parse: {lineno}:{column}: {faulty_line}") + exc = InvalidInput(f"Cannot parse: {lineno}:{column}: {faulty_line}") else: raise exc from None @@ -466,9 +1022,6 @@ def lib2to3_unparse(node: Node) -> str: return code -T = TypeVar("T") - - class Visitor(Generic[T]): """Basic lib2to3 visitor that yields things of type `T` on `visit()`.""" @@ -485,8 +1038,16 @@ class Visitor(Generic[T]): if node.type < 256: name = token.tok_name[node.type] else: - name = type_repr(node.type) - yield from getattr(self, f"visit_{name}", self.visit_default)(node) + name = str(type_repr(node.type)) + # We explicitly branch on whether a visitor exists (instead of + # using self.visit_default as the default arg to getattr) in order + # to save needing to create a bound method object and so mypyc can + # generate a native call to visit_default. + visitf = getattr(self, f"visit_{name}", None) + if visitf: + yield from visitf(node) + else: + yield from self.visit_default(node) def visit_default(self, node: LN) -> Iterator[T]: """Default `visit_*()` implementation. Recurses to children of `node`.""" @@ -520,19 +1081,19 @@ 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) -WHITESPACE = {token.DEDENT, token.INDENT, token.NEWLINE} -FLOW_CONTROL = {"return", "raise", "break", "continue"} -STATEMENT = { +WHITESPACE: Final = {token.DEDENT, token.INDENT, token.NEWLINE} +STATEMENT: Final = { syms.if_stmt, syms.while_stmt, syms.for_stmt, @@ -542,9 +1103,10 @@ STATEMENT = { syms.funcdef, syms.classdef, } -STANDALONE_COMMENT = 153 -LOGIC_OPERATORS = {"and", "or"} -COMPARATORS = { +STANDALONE_COMMENT: Final = 153 +token.tok_name[STANDALONE_COMMENT] = "STANDALONE_COMMENT" +LOGIC_OPERATORS: Final = {"and", "or"} +COMPARATORS: Final = { token.LESS, token.GREATER, token.EQEQUAL, @@ -552,7 +1114,7 @@ COMPARATORS = { token.LESSEQUAL, token.GREATEREQUAL, } -MATH_OPERATORS = { +MATH_OPERATORS: Final = { token.VBAR, token.CIRCUMFLEX, token.AMPER, @@ -568,21 +1130,23 @@ MATH_OPERATORS = { token.TILDE, token.DOUBLESTAR, } -STARS = {token.STAR, token.DOUBLESTAR} -VARARGS_PARENTS = { +STARS: Final = {token.STAR, token.DOUBLESTAR} +VARARGS_SPECIALS: Final = STARS | {token.SLASH} +VARARGS_PARENTS: Final = { syms.arglist, syms.argument, # double star in arglist syms.trailer, # single argument to call syms.typedargslist, syms.varargslist, # lambdas } -UNPACKING_PARENTS = { +UNPACKING_PARENTS: Final = { syms.atom, # single element of a list or set literal syms.dictsetmaker, syms.listmaker, syms.testlist_gexp, + syms.testlist_star_expr, } -TEST_DESCENDANTS = { +TEST_DESCENDANTS: Final = { syms.test, syms.lambdef, syms.or_test, @@ -599,7 +1163,7 @@ TEST_DESCENDANTS = { syms.term, syms.power, } -ASSIGNMENTS = { +ASSIGNMENTS: Final = { "=", "+=", "-=", @@ -615,28 +1179,29 @@ ASSIGNMENTS = { "**=", "//=", } -COMPREHENSION_PRIORITY = 20 -COMMA_PRIORITY = 18 -TERNARY_PRIORITY = 16 -LOGIC_PRIORITY = 14 -STRING_PRIORITY = 12 -COMPARATOR_PRIORITY = 10 -MATH_PRIORITIES = { - token.VBAR: 8, - token.CIRCUMFLEX: 7, - token.AMPER: 6, - token.LEFTSHIFT: 5, - token.RIGHTSHIFT: 5, - token.PLUS: 4, - token.MINUS: 4, - token.STAR: 3, - token.SLASH: 3, - token.DOUBLESLASH: 3, - token.PERCENT: 3, - token.AT: 3, - token.TILDE: 2, - token.DOUBLESTAR: 1, +COMPREHENSION_PRIORITY: Final = 20 +COMMA_PRIORITY: Final = 18 +TERNARY_PRIORITY: Final = 16 +LOGIC_PRIORITY: Final = 14 +STRING_PRIORITY: Final = 12 +COMPARATOR_PRIORITY: Final = 10 +MATH_PRIORITIES: Final = { + token.VBAR: 9, + token.CIRCUMFLEX: 8, + token.AMPER: 7, + token.LEFTSHIFT: 6, + token.RIGHTSHIFT: 6, + token.PLUS: 5, + token.MINUS: 5, + token.STAR: 4, + token.SLASH: 4, + token.DOUBLESLASH: 4, + token.PERCENT: 4, + token.AT: 4, + token.TILDE: 3, + token.DOUBLESTAR: 2, } +DOT_PRIORITY: Final = 1 @dataclass @@ -644,11 +1209,11 @@ class BracketTracker: """Keeps track of brackets on a line.""" depth: int = 0 - bracket_match: Dict[Tuple[Depth, NodeType], Leaf] = Factory(dict) - delimiters: Dict[LeafID, Priority] = Factory(dict) + bracket_match: Dict[Tuple[Depth, NodeType], Leaf] = field(default_factory=dict) + delimiters: Dict[LeafID, Priority] = field(default_factory=dict) previous: Optional[Leaf] = None - _for_loop_variable: int = 0 - _lambda_arguments: int = 0 + _for_loop_depths: List[int] = field(default_factory=list) + _lambda_argument_depths: List[int] = field(default_factory=list) def mark(self, leaf: Leaf) -> None: """Mark `leaf` with bracket-related metadata. Keep track of delimiters. @@ -694,7 +1259,7 @@ 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. @@ -702,6 +1267,17 @@ class BracketTracker: """ return max(v for k, v in self.delimiters.items() if k not in exclude) + 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. + """ + if not self.delimiters: + return 0 + + priority = priority or self.max_delimiter_priority() + return sum(1 for p in self.delimiters.values() if p == priority) + def maybe_increment_for_loop_variable(self, leaf: Leaf) -> bool: """In a for loop, or comprehension, the variables are often unpacks. @@ -710,16 +1286,21 @@ class BracketTracker: """ if leaf.type == token.NAME and leaf.value == "for": self.depth += 1 - self._for_loop_variable += 1 + self._for_loop_depths.append(self.depth) 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": + if ( + self._for_loop_depths + and self._for_loop_depths[-1] == self.depth + and leaf.type == token.NAME + and leaf.value == "in" + ): self.depth -= 1 - self._for_loop_variable -= 1 + self._for_loop_depths.pop() return True return False @@ -732,16 +1313,20 @@ class BracketTracker: """ if leaf.type == token.NAME and leaf.value == "lambda": self.depth += 1 - self._lambda_arguments += 1 + self._lambda_argument_depths.append(self.depth) 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: + if ( + self._lambda_argument_depths + and self._lambda_argument_depths[-1] == self.depth + and leaf.type == token.COLON + ): self.depth -= 1 - self._lambda_arguments -= 1 + self._lambda_argument_depths.pop() return True return False @@ -756,10 +1341,12 @@ class Line: """Holds leaves and comments. Can be printed with `str(line)`.""" depth: int = 0 - leaves: List[Leaf] = Factory(list) - comments: List[Tuple[Index, Leaf]] = Factory(list) - bracket_tracker: BracketTracker = Factory(BracketTracker) + leaves: List[Leaf] = field(default_factory=list) + # keys ordered like `leaves` + comments: Dict[LeafID, List[Leaf]] = field(default_factory=dict) + bracket_tracker: BracketTracker = field(default_factory=BracketTracker) inside_brackets: bool = False + should_explode: bool = False def append(self, leaf: Leaf, preformatted: bool = False) -> None: """Add a new `leaf` to the end of the line. @@ -830,6 +1417,76 @@ class Line: and self.leaves[0].value == "class" ) + @property + def is_stub_class(self) -> bool: + """Is this line a class definition with a body consisting only of "..."?""" + return self.is_class and self.leaves[-3:] == [ + Leaf(token.DOT, ".") for _ in range(3) + ] + + @property + def is_collection_with_optional_trailing_comma(self) -> bool: + """Is this line a collection literal with a trailing comma that's optional? + + Note that the trailing comma in a 1-tuple is not optional. + """ + if not self.leaves or len(self.leaves) < 4: + return False + + # Look for and address a trailing colon. + if self.leaves[-1].type == token.COLON: + closer = self.leaves[-2] + close_index = -2 + else: + closer = self.leaves[-1] + close_index = -1 + if closer.type not in CLOSING_BRACKETS or self.inside_brackets: + return False + + if closer.type == token.RPAR: + # Tuples require an extra check, because if there's only + # one element in the tuple removing the comma unmakes the + # tuple. + # + # We also check for parens before looking for the trailing + # comma because in some cases (eg assigning a dict + # literal) the literal gets wrapped in temporary parens + # during parsing. This case is covered by the + # collections.py test data. + opener = closer.opening_bracket + for _open_index, leaf in enumerate(self.leaves): + if leaf is opener: + break + + else: + # Couldn't find the matching opening paren, play it safe. + return False + + commas = 0 + comma_depth = self.leaves[close_index - 1].bracket_depth + for leaf in self.leaves[_open_index + 1 : close_index]: + if leaf.bracket_depth == comma_depth and leaf.type == token.COMMA: + commas += 1 + if commas > 1: + # We haven't looked yet for the trailing comma because + # we might also have caught noop parens. + return self.leaves[close_index - 1].type == token.COMMA + + elif commas == 1: + return False # it's either a one-tuple or didn't have a trailing comma + + if self.leaves[close_index - 1].type in CLOSING_BRACKETS: + close_index -= 1 + closer = self.leaves[close_index] + if closer.type == token.RPAR: + # TODO: this is a gut feeling. Will we ever see this? + return False + + if self.leaves[close_index - 1].type != token.COMMA: + return False + + return True + @property def is_def(self) -> bool: """Is this a function definition? (Also returns True for async defs.)""" @@ -842,35 +1499,11 @@ class Line: second_leaf: Optional[Leaf] = self.leaves[1] except IndexError: second_leaf = None - return ( - (first_leaf.type == token.NAME and first_leaf.value == "def") - or ( - first_leaf.type == token.ASYNC - and second_leaf is not None - and second_leaf.type == token.NAME - and second_leaf.value == "def" - ) - ) - - @property - def is_flow_control(self) -> bool: - """Is this line a flow control statement? - - Those are `return`, `raise`, `break`, and `continue`. - """ - return ( - bool(self) - and self.leaves[0].type == token.NAME - and self.leaves[0].value in FLOW_CONTROL - ) - - @property - def is_yield(self) -> bool: - """Is this line a yield statement?""" - return ( - bool(self) - and self.leaves[0].type == token.NAME - and self.leaves[0].value == "yield" + return (first_leaf.type == token.NAME and first_leaf.value == "def") or ( + first_leaf.type == token.ASYNC + and second_leaf is not None + and second_leaf.type == token.NAME + and second_leaf.value == "def" ) @property @@ -889,130 +1522,202 @@ class Line: and self.leaves[3].value == ")" ) + @property + def is_triple_quoted_string(self) -> bool: + """Is the line a triple quoted string?""" + return ( + bool(self) + and self.leaves[0].type == token.STRING + and self.leaves[0].value.startswith(('"""', "'''")) + ) + 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: - if leaf.bracket_depth <= depth_limit: - return True + if leaf.type == STANDALONE_COMMENT and leaf.bracket_depth <= depth_limit: + return True + + return False + + def contains_uncollapsable_type_comments(self) -> bool: + ignored_ids = set() + try: + last_leaf = self.leaves[-1] + ignored_ids.add(id(last_leaf)) + 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: + return False + + # A type comment is uncollapsable if it is attached to a leaf + # that isn't at the end of the line (since that could cause it + # to get associated to a different argument) or if there are + # comments before it (since that could cause it to get hidden + # behind a comment. + comment_seen = False + for leaf_id, comments in self.comments.items(): + for comment in comments: + if is_type_comment(comment): + if comment_seen or ( + not is_type_comment(comment, " ignore") + and leaf_id not in ignored_ids + ): + return True + + comment_seen = True + + return False + + def contains_unsplittable_type_ignore(self) -> bool: + if not self.leaves: + return False + + # If a 'type: ignore' is attached to the end of a line, we + # can't split the line, because we can't know which of the + # subexpressions the ignore was meant to apply to. + # + # We only want this to apply to actual physical lines from the + # original source, though: we don't want the presence of a + # 'type: ignore' at the end of a multiline expression to + # justify pushing it all onto one line. Thus we + # (unfortunately) need to check the actual source lines and + # only report an unsplittable 'type: ignore' if this line was + # one line in the original code. + + # Grab the first and last line numbers, skipping generated leaves + first_line = next((l.lineno for l in self.leaves if l.lineno != 0), 0) + last_line = next((l.lineno for l in reversed(self.leaves) if l.lineno != 0), 0) + + if first_line == last_line: + # We look at the last two leaves since a comma or an + # invisible paren could have been added at the end of the + # line. + for node in self.leaves[-2:]: + for comment in self.comments.get(id(node), []): + if is_type_comment(comment, " ignore"): + return True return False + def contains_multiline_strings(self) -> bool: + return any(is_multiline_string(leaf) for leaf in self.leaves) + def maybe_remove_trailing_comma(self, closing: Leaf) -> bool: """Remove trailing comma if there is one and it's safe.""" + if not (self.leaves and self.leaves[-1].type == token.COMMA): + return False + + # We remove trailing commas only in the case of importing a + # single name from a module. if not ( self.leaves + and self.is_import + and len(self.leaves) > 4 and self.leaves[-1].type == token.COMMA and closing.type in CLOSING_BRACKETS + and self.leaves[-4].type == token.NAME + and ( + # regular `from foo import bar,` + self.leaves[-4].value == "import" + # `from foo import (bar as baz,) + or ( + len(self.leaves) > 6 + and self.leaves[-6].value == "import" + and self.leaves[-3].value == "as" + ) + # `from foo import bar as baz,` + or ( + len(self.leaves) > 5 + and self.leaves[-5].value == "import" + and self.leaves[-3].value == "as" + ) + ) + and closing.type == token.RPAR ): return False - if closing.type == token.RBRACE: - self.remove_trailing_comma() - return True + self.remove_trailing_comma() + return True - if closing.type == token.RSQB: - comma = self.leaves[-1] - if comma.parent and comma.parent.type == syms.listmaker: - self.remove_trailing_comma() - return True - - # For parens let's check if it's safe to remove the comma. - # Imports are always safe. - if self.is_import: - self.remove_trailing_comma() - return True - - # Otheriwsse, 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 - opening = closing.opening_bracket - for _opening_index, leaf in enumerate(self.leaves): - if leaf is opening: - break - - else: - return False - - for leaf in self.leaves[_opening_index + 1 :]: - if leaf is closing: - break - - 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: - commas += 1 - break - - if commas > 1: - self.remove_trailing_comma() - return True - - return False - - def append_comment(self, comment: Leaf) -> bool: - """Add an inline or standalone comment to the line.""" - if ( - comment.type == STANDALONE_COMMENT - and self.bracket_tracker.any_open_brackets() - ): - comment.prefix = "" - return False + def append_comment(self, comment: Leaf) -> bool: + """Add an inline or standalone comment to the line.""" + if ( + comment.type == STANDALONE_COMMENT + and self.bracket_tracker.any_open_brackets() + ): + comment.prefix = "" + return False if comment.type != token.COMMENT: return False - after = len(self.leaves) - 1 - if after == -1: + if not self.leaves: comment.type = STANDALONE_COMMENT comment.prefix = "" return False - else: - self.comments.append((after, comment)) - return True + 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) -> Iterator[Leaf]: + def comments_after(self, leaf: Leaf) -> List[Leaf]: """Generate comments that should appear directly after `leaf`.""" - for _leaf_index, _leaf in enumerate(self.leaves): - if leaf is _leaf: - break - - else: - return - - for index, comment_after in self.comments: - if _leaf_index == index: - yield comment_after + return self.comments.get(id(leaf), []) def remove_trailing_comma(self) -> None: """Remove the trailing comma and moves the comments attached to it.""" - comma_index = len(self.leaves) - 1 - for i in range(len(self.comments)): - comment_index, comment = self.comments[i] - if comment_index == comma_index: - self.comments[i] = (comma_index - 1, comment) - self.leaves.pop() + trailing_comma = self.leaves.pop() + trailing_comma_comments = self.comments.pop(id(trailing_comma), []) + self.comments.setdefault(id(self.leaves[-1]), []).extend( + trailing_comma_comments + ) 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) - return ( - subscript_start is not None - and any(n.type in TEST_DESCENDANTS for n in subscript_start.pre_order()) + + 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() + ) + + def clone(self) -> "Line": + return Line( + depth=self.depth, + inside_brackets=self.inside_brackets, + should_explode=self.should_explode, ) def __str__(self) -> str: @@ -1026,8 +1731,9 @@ class Line: res = f"{first.prefix}{indent}{first.value}" for leaf in leaves: res += str(leaf) - for _, comment in self.comments: + for comment in itertools.chain.from_iterable(self.comments.values()): res += str(comment) + return res + "\n" def __bool__(self) -> bool: @@ -1035,55 +1741,6 @@ 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 @@ -1093,22 +1750,26 @@ class EmptyLineTracker: the prefix of the first leaf consists of optional newlines. Those newlines are consumed by `maybe_empty_lines()` and included in the computation. """ + + is_pyi: bool = False previous_line: Optional[Line] = None previous_after: int = 0 - previous_defs: List[int] = Factory(list) + previous_defs: List[int] = field(default_factory=list) def maybe_empty_lines(self, current_line: Line) -> Tuple[int, int]: """Return the number of extra empty lines before and after the `current_line`. This is for separating `def`, `async def` and `class` with extra empty - lines (two on module-level), as well as providing an extra empty line - after flow control keywords to make them more prominent. + 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 + before = ( + # Black should not insert empty lines at the beginning + # of the file + 0 + if self.previous_line is None + else before - self.previous_after + ) self.previous_after = after self.previous_line = current_line return before, after @@ -1116,7 +1777,7 @@ class EmptyLineTracker: def _maybe_empty_lines(self, current_line: Line) -> Tuple[int, int]: max_allowed = 1 if current_line.depth == 0: - max_allowed = 2 + max_allowed = 1 if self.is_pyi else 2 if current_line.leaves: # Consume the first leaf's extra newlines. first_leaf = current_line.leaves[0] @@ -1128,29 +1789,12 @@ class EmptyLineTracker: depth = current_line.depth while self.previous_defs and self.previous_defs[-1] >= depth: self.previous_defs.pop() - 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.is_comment - and self.previous_line.depth == current_line.depth - and before == 0 - ): - return 0, 0 - - newlines = 2 - if current_line.depth: - newlines -= 1 - return newlines, 0 + if self.is_pyi: + before = 0 if depth else 1 + else: + before = 1 if depth else 2 + 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 @@ -1160,8 +1804,59 @@ class EmptyLineTracker: ): return (before or 1), 0 + if ( + self.previous_line + and self.previous_line.is_class + and current_line.is_triple_quoted_string + ): + return before, 1 + 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]): @@ -1170,9 +1865,13 @@ class LineGenerator(Visitor[Line]): Note: destroys the tree it's visiting by mutating prefixes of its leaves in ways that will no longer stringify to valid Python code on the tree. """ - current_line: Line = Factory(Line) - def line(self, indent: int = 0, type: Type[Line] = Line) -> Iterator[Line]: + is_pyi: bool = False + normalize_strings: bool = True + current_line: Line = field(default_factory=Line) + remove_u_prefix: bool = False + + def line(self, indent: int = 0) -> Iterator[Line]: """Generate a line. If the line is empty, only emit if it makes sense. @@ -1181,75 +1880,50 @@ 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 node.type == token.STRING: - 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) + 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]: + def visit_INDENT(self, node: Leaf) -> Iterator[Line]: """Increase indentation level, maybe yield a line.""" # In blib2to3 INDENT never holds comments. yield from self.line(+1) yield from self.visit_default(node) - def visit_DEDENT(self, node: Node) -> Iterator[Line]: + def visit_DEDENT(self, node: Leaf) -> Iterator[Line]: """Decrease indentation level, maybe yield a line.""" # The current line might still wait for trailing comments. At DEDENT time # there won't be any (they would be prefixes on the preceding NEWLINE). @@ -1274,7 +1948,7 @@ class LineGenerator(Visitor[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 a set of string leaf values immeditely after which + `parens` holds a set of string leaf values immediately after which invisible parens should be put. """ normalize_invisible_parens(node, parens_after=parens) @@ -1284,16 +1958,27 @@ class LineGenerator(Visitor[Line]): yield from self.visit(child) + def visit_suite(self, node: Node) -> Iterator[Line]: + """Visit a suite.""" + if self.is_pyi and is_stub_suite(node): + yield from self.visit(node.children[2]) + else: + yield from self.visit_default(node) + def visit_simple_stmt(self, node: Node) -> Iterator[Line]: """Visit a statement without nested statements.""" is_suite_like = node.parent and node.parent.type in STATEMENT if is_suite_like: - yield from self.line(+1) - yield from self.visit_default(node) - yield from self.line(-1) + if self.is_pyi and is_stub_body(node): + yield from self.visit_default(node) + else: + yield from self.line(+1) + yield from self.visit_default(node) + yield from self.line(-1) else: - yield from self.line() + if not self.is_pyi or not node.parent or not is_stub_suite(node.parent): + yield from self.line() yield from self.visit_default(node) def visit_async_stmt(self, node: Node) -> Iterator[Line]: @@ -1317,32 +2002,6 @@ 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() @@ -1352,25 +2011,41 @@ 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) + 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) - 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) + def visit_factor(self, node: Node) -> Iterator[Line]: + """Force parentheses between a unary op and a binary power: - if node.type == token.ENDMARKER: - # somebody decided not to put a final `# fmt: on` - yield from self.line() + -2 ** 8 -> -(2 ** 8) + """ + _operator, operand = node.children + if ( + operand.type == syms.power + and len(operand.children) == 3 + and operand.children[1].type == token.DOUBLESTAR + ): + lpar = Leaf(token.LPAR, "(") + rpar = Leaf(token.RPAR, ")") + index = operand.remove() or 0 + node.insert_child(index, Node(syms.atom, [lpar, operand, rpar])) + yield from self.visit_default(node) - def __attrs_post_init__(self) -> None: + def visit_STRING(self, leaf: Leaf) -> Iterator[Line]: + # Check if it's a docstring + if prev_siblings_are( + leaf.parent, [None, token.NEWLINE, token.INDENT, syms.simple_stmt] + ) and is_multiline_string(leaf): + prefix = " " * self.current_line.depth + docstring = fix_docstring(leaf.value[3:-3], prefix) + leaf.value = leaf.value[0:3] + docstring + leaf.value[-3:] + normalize_string_quotes(leaf) + + yield from self.visit_default(leaf) + + def __post_init__(self) -> None: """You are in a twisty little maze of passages.""" v = self.visit_stmt Ø: Set[str] = set() @@ -1389,6 +2064,8 @@ class LineGenerator(Visitor[Line]): self.visit_classdef = partial(v, keywords={"class"}, parens=Ø) self.visit_expr_stmt = partial(v, keywords=Ø, parens=ASSIGNMENTS) self.visit_return_stmt = partial(v, keywords={"return"}, parens={"return"}) + self.visit_import_from = partial(v, keywords=Ø, parens={"import"}) + self.visit_del_stmt = partial(v, keywords=Ø, parens={"del"}) self.visit_async_funcdef = self.visit_async_stmt self.visit_decorated = self.visit_decorators @@ -1401,7 +2078,7 @@ BRACKETS = OPENING_BRACKETS | CLOSING_BRACKETS ALWAYS_NO_SPACE = CLOSING_BRACKETS | {token.COMMA, STANDALONE_COMMENT} -def whitespace(leaf: Leaf, *, complex_subscript: bool) -> str: # noqa C901 +def whitespace(leaf: Leaf, *, complex_subscript: bool) -> str: # noqa: C901 """Return whitespace prefix if needed for the given `leaf`. `complex_subscript` signals whether the given leaf is part of a subscription @@ -1420,10 +2097,11 @@ def whitespace(leaf: Leaf, *, complex_subscript: bool) -> str: # noqa C901 return DOUBLESPACE assert p is not None, f"INTERNAL ERROR: hand-made leaf without parent: {leaf!r}" - if ( - t == token.COLON - and p.type not in {syms.subscript, syms.subscriptlist, syms.sliceop} - ): + if t == token.COLON and p.type not in { + syms.subscript, + syms.subscriptlist, + syms.sliceop, + }: return NO prev = leaf.prev_sibling @@ -1444,7 +2122,10 @@ def whitespace(leaf: Leaf, *, complex_subscript: bool) -> str: # noqa C901 if prevp.type == token.EQUAL: if prevp.parent: if prevp.parent.type in { - syms.arglist, syms.argument, syms.parameters, syms.varargslist + syms.arglist, + syms.argument, + syms.parameters, + syms.varargslist, }: return NO @@ -1454,7 +2135,7 @@ def whitespace(leaf: Leaf, *, complex_subscript: bool) -> str: # noqa C901 # that, too. return prevp.prefix - elif prevp.type in STARS: + elif prevp.type in VARARGS_SPECIALS: if is_vararg(prevp, within=VARARGS_PARENTS | UNPACKING_PARENTS): return NO @@ -1544,7 +2225,7 @@ def whitespace(leaf: Leaf, *, complex_subscript: bool) -> str: # noqa C901 if not prevp or prevp.type == token.LPAR: return NO - elif prev.type in {token.EQUAL} | STARS: + elif prev.type in {token.EQUAL} | VARARGS_SPECIALS: return NO elif p.type == syms.decorator: @@ -1597,16 +2278,16 @@ def whitespace(leaf: Leaf, *, complex_subscript: bool) -> 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: 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: @@ -1645,6 +2326,22 @@ def preceding_leaf(node: Optional[LN]) -> Optional[Leaf]: return None +def prev_siblings_are(node: Optional[LN], tokens: List[Optional[NodeType]]) -> bool: + """Return if the `node` and its previous siblings match types against the provided + list of tokens; the provided `node`has its type matched against the last element in + the list. `None` can be used as the first element to declare that the start of the + list is anchored at the start of its parent's children.""" + if not tokens: + return True + if tokens[-1] is None: + return node is None + if not node: + return False + if node.type != tokens[-1]: + return False + return prev_siblings_are(node.prev_sibling, tokens[:-1]) + + def child_towards(ancestor: Node, descendant: LN) -> Optional[LN]: """Return the child of `ancestor` that contains `descendant`.""" node: Optional[LN] = descendant @@ -1653,7 +2350,32 @@ def child_towards(ancestor: Node, descendant: LN) -> Optional[LN]: return node -def is_split_after_delimiter(leaf: Leaf, previous: Leaf = None) -> int: +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: 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 @@ -1667,8 +2389,8 @@ def is_split_after_delimiter(leaf: Leaf, previous: Leaf = None) -> int: return 0 -def is_split_before_delimiter(leaf: Leaf, previous: Leaf = None) -> int: - """Return the priority of the `leaf` delimiter, given a line before after it. +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 cause a line break before themselves. @@ -1680,6 +2402,14 @@ def is_split_before_delimiter(leaf: Leaf, previous: Leaf = None) -> int: # Don't treat them as a delimiter. return 0 + if ( + leaf.type == token.DOT + and leaf.parent + and leaf.parent.type not in {syms.import_from, syms.dotted_name} + and (previous is None or previous.type in CLOSING_BRACKETS) + ): + return DOT_PRIORITY + if ( leaf.type in MATH_OPERATORS and leaf.parent @@ -1697,37 +2427,69 @@ def is_split_before_delimiter(leaf: Leaf, previous: Leaf = None) -> int: ): return STRING_PRIORITY + if leaf.type not in {token.NAME, token.ASYNC}: + return 0 + if ( - leaf.type == token.NAME - and leaf.value == "for" + leaf.value == "for" and leaf.parent and leaf.parent.type in {syms.comp_for, syms.old_comp_for} + or leaf.type == token.ASYNC ): - return COMPREHENSION_PRIORITY + if ( + not isinstance(leaf.prev_sibling, Leaf) + or leaf.prev_sibling.value != "async" + ): + return COMPREHENSION_PRIORITY if ( - leaf.type == token.NAME - and leaf.value == "if" + leaf.value == "if" and leaf.parent and leaf.parent.type in {syms.comp_if, syms.old_comp_if} ): return COMPREHENSION_PRIORITY + if leaf.value in {"if", "else"} and leaf.parent and leaf.parent.type == syms.test: + return TERNARY_PRIORITY + + if leaf.value == "is": + return COMPARATOR_PRIORITY + if ( - leaf.type == token.NAME - and leaf.value in {"if", "else"} + leaf.value == "in" and leaf.parent - and leaf.parent.type == syms.test + and leaf.parent.type in {syms.comp_op, syms.comparison} + and not ( + previous is not None + and previous.type == token.NAME + and previous.value == "not" + ) ): - return TERNARY_PRIORITY + return COMPARATOR_PRIORITY + + if ( + leaf.value == "not" + and leaf.parent + and leaf.parent.type == syms.comp_op + and not ( + previous is not None + and previous.type == token.NAME + and previous.value == "is" + ) + ): + return COMPARATOR_PRIORITY - if leaf.type == token.NAME and leaf.value in LOGIC_OPERATORS and leaf.parent: + if leaf.value in LOGIC_OPERATORS and leaf.parent: return LOGIC_PRIORITY return 0 -def generate_comments(leaf: Leaf) -> Iterator[Leaf]: +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. Comments in lib2to3 are shoved into the whitespace prefix. This happens @@ -1746,49 +2508,70 @@ def generate_comments(leaf: Leaf) -> 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: + """Describes a piece of syntax that is a comment. + + It's not a :class:`blib2to3.pytree.Leaf` so that: + + * it can be cached (`Leaf` objects should not be reused more than once as + they store their lineno, column, prefix, and parent information); + * `newlines` and `consumed` fields are kept separate from the `value`. This + simplifies handling of special marker comments like ``# fmt: off/on``. + """ + + 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]: + """Return a list of :class:`ProtoComment` objects parsed from the given `prefix`.""" + result: List[ProtoComment] = [] + if not prefix or "#" not in prefix: + return result consumed = 0 nlines = 0 - for index, line in enumerate(p.split("\n")): + 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 leaf.type != token.ENDMARKER: + if index == ignored_lines 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: """Return a consistently formatted comment from the given `content` string. - All comments (except for "##", "#!", "#:") should have a single space between - the hash sign and the content. + All comments (except for "##", "#!", "#:", '#'", "#%%") should have a single + space between the hash sign and the content. If `content` didn't start with a hash sign, one is provided. """ @@ -1798,62 +2581,110 @@ def make_comment(content: str) -> str: if content[0] == "#": content = content[1:] - if content and content[0] not in " !:#": + if content and content[0] not in " !:#'%": content = " " + content return "#" + content -def split_line( - line: Line, line_length: int, inner: bool = False, py36: bool = False +def transform_line( + line: Line, + line_length: int, + normalize_strings: bool, + features: Collection[Feature] = (), ) -> Iterator[Line]: - """Split a `line` into potentially many lines. + """Transform a `line`, potentially splitting it into many lines. They should fit in the allotted `line_length` but might not be able to. - `inner` signifies that there were a pair of brackets somewhere around the - current `line`, possibly transitively. This means we can fallback to splitting - by delimiters if the LHS/RHS don't yield any results. - If `py36` is True, splitting may generate syntax that is only compatible - with Python 3.6 and later. + `features` are syntactical features that may be used in the output. """ - if isinstance(line, UnformattedLines) or line.is_comment: + if line.is_comment: yield line return - line_str = str(line).strip("\n") + line_str = line_to_string(line) + + def init_st(ST: Type[StringTransformer]) -> StringTransformer: + """Initialize StringTransformer""" + return ST(line_length, normalize_strings) + + string_merge = init_st(StringMerger) + string_paren_strip = init_st(StringParenStripper) + string_split = init_st(StringSplitter) + string_paren_wrap = init_st(StringParenWrapper) + + transformers: List[Transformer] if ( - len(line_str) <= line_length - and "\n" not in line_str # multiline strings - and not line.contains_standalone_comments() + not line.contains_uncollapsable_type_comments() + and not line.should_explode + and not line.is_collection_with_optional_trailing_comma + and ( + is_line_short_enough(line, line_length=line_length, line_str=line_str) + or line.contains_unsplittable_type_ignore() + ) + and not (line.contains_standalone_comments() and line.inside_brackets) ): - yield line - return - - split_funcs: List[SplitFunc] - if line.is_def: - split_funcs = [left_hand_split] - elif line.is_import: - split_funcs = [explode_split] - elif line.inside_brackets: - split_funcs = [delimiter_split, standalone_comment_split, right_hand_split] + # Only apply basic string preprocessing, since lines shouldn't be split here. + transformers = [string_merge, string_paren_strip] + elif line.is_def: + transformers = [left_hand_split] else: - split_funcs = [right_hand_split] - for split_func in split_funcs: + + def rhs(line: Line, features: Collection[Feature]) -> Iterator[Line]: + for omit in generate_trailers_to_omit(line, line_length): + lines = list(right_hand_split(line, line_length, features, omit=omit)) + if is_line_short_enough(lines[0], line_length=line_length): + yield from lines + return + + # All splits failed, best effort split with no omits. + # This mostly happens to multiline strings that are by definition + # reported as not fitting a single line. + # line_length=1 here was historically a bug that somehow became a feature. + # See #762 and #781 for the full story. + yield from right_hand_split(line, line_length=1, features=features) + + if line.inside_brackets: + transformers = [ + string_merge, + string_paren_strip, + delimiter_split, + standalone_comment_split, + string_split, + string_paren_wrap, + rhs, + ] + else: + transformers = [ + string_merge, + string_paren_strip, + string_split, + string_paren_wrap, + rhs, + ] + + for transform in transformers: # We are accumulating lines in `result` because we might want to abort # mission and return the original line in the end, or attempt a different # split altogether. result: List[Line] = [] try: - for l in split_func(line, py36): + for l in transform(line, features): if str(l).strip("\n") == line_str: - raise CannotSplit("Split function returned an unchanged result") + raise CannotTransform( + "Line transformer returned an unchanged result" + ) result.extend( - split_line(l, line_length=line_length, inner=True, py36=py36) + transform_line( + l, + line_length=line_length, + normalize_strings=normalize_strings, + features=features, + ) ) - except CannotSplit as cs: + except CannotTransform: continue - else: yield from result break @@ -1862,42 +2693,1967 @@ def split_line( yield line -def left_hand_split(line: Line, py36: bool = False) -> Iterator[Line]: - """Split line into many lines, starting with the first matching bracket pair. +@dataclass # type: ignore +class StringTransformer(ABC): + """ + An implementation of the Transformer protocol that relies on its + subclasses overriding the template methods `do_match(...)` and + `do_transform(...)`. - Note: this usually looks weird, only use this for function definitions. - Prefer RHS otherwise. This is why this function is not symmetrical with - :func:`right_hand_split` which also handles optional parentheses. + This Transformer works exclusively on strings (for example, by merging + or splitting them). + + The following sections can be found among the docstrings of each concrete + StringTransformer subclass. + + Requirements: + Which requirements must be met of the given Line for this + StringTransformer to be applied? + + Transformations: + If the given Line meets all of the above requirments, which string + transformations can you expect to be applied to it by this + StringTransformer? + + Collaborations: + What contractual agreements does this StringTransformer have with other + StringTransfomers? Such collaborations should be eliminated/minimized + as much as possible. """ - head = Line(depth=line.depth) - body = Line(depth=line.depth + 1, inside_brackets=True) - tail = Line(depth=line.depth) - tail_leaves: List[Leaf] = [] - body_leaves: List[Leaf] = [] - head_leaves: List[Leaf] = [] - current_leaves = head_leaves - matching_bracket = None - for leaf in line.leaves: - if ( - current_leaves is body_leaves - and leaf.type in CLOSING_BRACKETS - and leaf.opening_bracket is matching_bracket - ): - current_leaves = tail_leaves if body_leaves else head_leaves + + line_length: int + normalize_strings: bool + + @abstractmethod + def do_match(self, line: Line) -> TMatchResult: + """ + Returns: + * Ok(string_idx) such that `line.leaves[string_idx]` is our target + string, if a match was able to be made. + OR + * Err(CannotTransform), if a match was not able to be made. + """ + + @abstractmethod + def do_transform(self, line: Line, string_idx: int) -> Iterator[TResult[Line]]: + """ + Yields: + * Ok(new_line) where new_line is the new transformed line. + OR + * Err(CannotTransform) if the transformation failed for some reason. The + `do_match(...)` template method should usually be used to reject + the form of the given Line, but in some cases it is difficult to + know whether or not a Line meets the StringTransformer's + requirements until the transformation is already midway. + + Side Effects: + This method should NOT mutate @line directly, but it MAY mutate the + Line's underlying Node structure. (WARNING: If the underlying Node + structure IS altered, then this method should NOT be allowed to + yield an CannotTransform after that point.) + """ + + def __call__(self, line: Line, _features: Collection[Feature]) -> Iterator[Line]: + """ + StringTransformer instances have a call signature that mirrors that of + the Transformer type. + + Raises: + CannotTransform(...) if the concrete StringTransformer class is unable + to transform @line. + """ + # Optimization to avoid calling `self.do_match(...)` when the line does + # not contain any string. + if not any(leaf.type == token.STRING for leaf in line.leaves): + raise CannotTransform("There are no strings in this line.") + + match_result = self.do_match(line) + + if isinstance(match_result, Err): + cant_transform = match_result.err() + raise CannotTransform( + f"The string transformer {self.__class__.__name__} does not recognize" + " this line as one that it can transform." + ) from cant_transform + + string_idx = match_result.ok() + + for line_result in self.do_transform(line, string_idx): + if isinstance(line_result, Err): + cant_transform = line_result.err() + raise CannotTransform( + "StringTransformer failed while attempting to transform string." + ) from cant_transform + line = line_result.ok() + yield line + + +@dataclass +class CustomSplit: + """A custom (i.e. manual) string split. + + A single CustomSplit instance represents a single substring. + + Examples: + Consider the following string: + ``` + "Hi there friend." + " This is a custom" + f" string {split}." + ``` + + This string will correspond to the following three CustomSplit instances: + ``` + CustomSplit(False, 16) + CustomSplit(False, 17) + CustomSplit(True, 16) + ``` + """ + + has_prefix: bool + break_idx: int + + +class CustomSplitMapMixin: + """ + This mixin class is used to map merged strings to a sequence of + CustomSplits, which will then be used to re-split the strings iff none of + the resultant substrings go over the configured max line length. + """ + + _Key = Tuple[StringID, str] + _CUSTOM_SPLIT_MAP: Dict[_Key, Tuple[CustomSplit, ...]] = defaultdict(tuple) + + @staticmethod + def _get_key(string: str) -> "CustomSplitMapMixin._Key": + """ + Returns: + A unique identifier that is used internally to map @string to a + group of custom splits. + """ + return (id(string), string) + + def add_custom_splits( + self, string: str, custom_splits: Iterable[CustomSplit] + ) -> None: + """Custom Split Map Setter Method + + Side Effects: + Adds a mapping from @string to the custom splits @custom_splits. + """ + key = self._get_key(string) + self._CUSTOM_SPLIT_MAP[key] = tuple(custom_splits) + + def pop_custom_splits(self, string: str) -> List[CustomSplit]: + """Custom Split Map Getter Method + + Returns: + * A list of the custom splits that are mapped to @string, if any + exist. + OR + * [], otherwise. + + Side Effects: + Deletes the mapping between @string and its associated custom + splits (which are returned to the caller). + """ + key = self._get_key(string) + + custom_splits = self._CUSTOM_SPLIT_MAP[key] + del self._CUSTOM_SPLIT_MAP[key] + + return list(custom_splits) + + def has_custom_splits(self, string: str) -> bool: + """ + Returns: + True iff @string is associated with a set of custom splits. + """ + key = self._get_key(string) + return key in self._CUSTOM_SPLIT_MAP + + +class StringMerger(CustomSplitMapMixin, StringTransformer): + """StringTransformer that merges strings together. + + Requirements: + (A) The line contains adjacent strings such that at most one substring + has inline comments AND none of those inline comments are pragmas AND + the set of all substring prefixes is either of length 1 or equal to + {"", "f"} AND none of the substrings are raw strings (i.e. are prefixed + with 'r'). + OR + (B) The line contains a string which uses line continuation backslashes. + + Transformations: + Depending on which of the two requirements above where met, either: + + (A) The string group associated with the target string is merged. + OR + (B) All line-continuation backslashes are removed from the target string. + + Collaborations: + StringMerger provides custom split information to StringSplitter. + """ + + def do_match(self, line: Line) -> TMatchResult: + LL = line.leaves + + is_valid_index = is_valid_index_factory(LL) + + for (i, leaf) in enumerate(LL): + if ( + leaf.type == token.STRING + and is_valid_index(i + 1) + and LL[i + 1].type == token.STRING + ): + return Ok(i) + + if leaf.type == token.STRING and "\\\n" in leaf.value: + return Ok(i) + + return TErr("This line has no strings that need merging.") + + def do_transform(self, line: Line, string_idx: int) -> Iterator[TResult[Line]]: + new_line = line + rblc_result = self.__remove_backslash_line_continuation_chars( + new_line, string_idx + ) + if isinstance(rblc_result, Ok): + new_line = rblc_result.ok() + + msg_result = self.__merge_string_group(new_line, string_idx) + if isinstance(msg_result, Ok): + new_line = msg_result.ok() + + if isinstance(rblc_result, Err) and isinstance(msg_result, Err): + msg_cant_transform = msg_result.err() + rblc_cant_transform = rblc_result.err() + cant_transform = CannotTransform( + "StringMerger failed to merge any strings in this line." + ) + + # Chain the errors together using `__cause__`. + msg_cant_transform.__cause__ = rblc_cant_transform + cant_transform.__cause__ = msg_cant_transform + + yield Err(cant_transform) + else: + yield Ok(new_line) + + @staticmethod + def __remove_backslash_line_continuation_chars( + line: Line, string_idx: int + ) -> TResult[Line]: + """ + Merge strings that were split across multiple lines using + line-continuation backslashes. + + Returns: + Ok(new_line), if @line contains backslash line-continuation + characters. + OR + Err(CannotTransform), otherwise. + """ + LL = line.leaves + + string_leaf = LL[string_idx] + if not ( + string_leaf.type == token.STRING + and "\\\n" in string_leaf.value + and not has_triple_quotes(string_leaf.value) + ): + return TErr( + f"String leaf {string_leaf} does not contain any backslash line" + " continuation characters." + ) + + new_line = line.clone() + new_line.comments = line.comments + append_leaves(new_line, line, LL) + + new_string_leaf = new_line.leaves[string_idx] + new_string_leaf.value = new_string_leaf.value.replace("\\\n", "") + + return Ok(new_line) + + def __merge_string_group(self, line: Line, string_idx: int) -> TResult[Line]: + """ + Merges string group (i.e. set of adjacent strings) where the first + string in the group is `line.leaves[string_idx]`. + + Returns: + Ok(new_line), if ALL of the validation checks found in + __validate_msg(...) pass. + OR + Err(CannotTransform), otherwise. + """ + LL = line.leaves + + is_valid_index = is_valid_index_factory(LL) + + vresult = self.__validate_msg(line, string_idx) + if isinstance(vresult, Err): + return vresult + + # If the string group is wrapped inside an Atom node, we must make sure + # to later replace that Atom with our new (merged) string leaf. + atom_node = LL[string_idx].parent + + # We will place BREAK_MARK in between every two substrings that we + # merge. We will then later go through our final result and use the + # various instances of BREAK_MARK we find to add the right values to + # the custom split map. + BREAK_MARK = "@@@@@ BLACK BREAKPOINT MARKER @@@@@" + + QUOTE = LL[string_idx].value[-1] + + def make_naked(string: str, string_prefix: str) -> str: + """Strip @string (i.e. make it a "naked" string) + + Pre-conditions: + * assert_is_leaf_string(@string) + + Returns: + A string that is identical to @string except that + @string_prefix has been stripped, the surrounding QUOTE + characters have been removed, and any remaining QUOTE + characters have been escaped. + """ + assert_is_leaf_string(string) + + RE_EVEN_BACKSLASHES = r"(?:(?= 0 + ), "Logic error while filling the custom string breakpoint cache." + + temp_string = temp_string[mark_idx + len(BREAK_MARK) :] + breakpoint_idx = mark_idx + (len(prefix) if has_prefix else 0) + 1 + custom_splits.append(CustomSplit(has_prefix, breakpoint_idx)) + + string_leaf = Leaf(token.STRING, S_leaf.value.replace(BREAK_MARK, "")) + + if atom_node is not None: + replace_child(atom_node, string_leaf) + + # Build the final line ('new_line') that this method will later return. + new_line = line.clone() + for (i, leaf) in enumerate(LL): + if i == string_idx: + new_line.append(string_leaf) + + if string_idx <= i < string_idx + num_of_strings: + for comment_leaf in line.comments_after(LL[i]): + new_line.append(comment_leaf, preformatted=True) + continue + + append_leaves(new_line, line, [leaf]) + + self.add_custom_splits(string_leaf.value, custom_splits) + return Ok(new_line) + + @staticmethod + def __validate_msg(line: Line, string_idx: int) -> TResult[None]: + """Validate (M)erge (S)tring (G)roup + + Transform-time string validation logic for __merge_string_group(...). + + Returns: + * Ok(None), if ALL validation checks (listed below) pass. + OR + * Err(CannotTransform), if any of the following are true: + - The target string is not in a string group (i.e. it has no + adjacent strings). + - The string group has more than one inline comment. + - The string group has an inline comment that appears to be a pragma. + - The set of all string prefixes in the string group is of + length greater than one and is not equal to {"", "f"}. + - The string group consists of raw strings. + """ + num_of_inline_string_comments = 0 + set_of_prefixes = set() + num_of_strings = 0 + for leaf in line.leaves[string_idx:]: + if leaf.type != token.STRING: + # If the string group is trailed by a comma, we count the + # comments trailing the comma to be one of the string group's + # comments. + if leaf.type == token.COMMA and id(leaf) in line.comments: + num_of_inline_string_comments += 1 + break + + if has_triple_quotes(leaf.value): + return TErr("StringMerger does NOT merge multiline strings.") + + num_of_strings += 1 + prefix = get_string_prefix(leaf.value) + if "r" in prefix: + return TErr("StringMerger does NOT merge raw strings.") + + set_of_prefixes.add(prefix) + + if id(leaf) in line.comments: + num_of_inline_string_comments += 1 + if contains_pragma_comment(line.comments[id(leaf)]): + return TErr("Cannot merge strings which have pragma comments.") + + if num_of_strings < 2: + return TErr( + f"Not enough strings to merge (num_of_strings={num_of_strings})." + ) + + if num_of_inline_string_comments > 1: + return TErr( + f"Too many inline string comments ({num_of_inline_string_comments})." + ) + + if len(set_of_prefixes) > 1 and set_of_prefixes != {"", "f"}: + return TErr(f"Too many different prefixes ({set_of_prefixes}).") + + return Ok(None) + + +class StringParenStripper(StringTransformer): + """StringTransformer that strips surrounding parentheses from strings. + + Requirements: + The line contains a string which is surrounded by parentheses and: + - The target string is NOT the only argument to a function call). + - The RPAR is NOT followed by an attribute access (i.e. a dot). + + Transformations: + The parentheses mentioned in the 'Requirements' section are stripped. + + Collaborations: + StringParenStripper has its own inherent usefulness, but it is also + relied on to clean up the parentheses created by StringParenWrapper (in + the event that they are no longer needed). + """ + + def do_match(self, line: Line) -> TMatchResult: + LL = line.leaves + + is_valid_index = is_valid_index_factory(LL) + + for (idx, leaf) in enumerate(LL): + # Should be a string... + if leaf.type != token.STRING: + continue + + # Should be preceded by a non-empty LPAR... + if ( + not is_valid_index(idx - 1) + or LL[idx - 1].type != token.LPAR + or is_empty_lpar(LL[idx - 1]) + ): + continue + + # That LPAR should NOT be preceded by a function name or a closing + # bracket (which could be a function which returns a function or a + # list/dictionary that contains a function)... + if is_valid_index(idx - 2) and ( + LL[idx - 2].type == token.NAME or LL[idx - 2].type in CLOSING_BRACKETS + ): + continue + + string_idx = idx + + # Skip the string trailer, if one exists. + string_parser = StringParser() + next_idx = string_parser.parse(LL, string_idx) + + # Should be followed by a non-empty RPAR... + if ( + is_valid_index(next_idx) + and LL[next_idx].type == token.RPAR + and not is_empty_rpar(LL[next_idx]) + ): + # That RPAR should NOT be followed by a '.' symbol. + if is_valid_index(next_idx + 1) and LL[next_idx + 1].type == token.DOT: + continue + + return Ok(string_idx) + + return TErr("This line has no strings wrapped in parens.") + + def do_transform(self, line: Line, string_idx: int) -> Iterator[TResult[Line]]: + LL = line.leaves + + string_parser = StringParser() + rpar_idx = string_parser.parse(LL, string_idx) + + for leaf in (LL[string_idx - 1], LL[rpar_idx]): + if line.comments_after(leaf): + yield TErr( + "Will not strip parentheses which have comments attached to them." + ) + + new_line = line.clone() + new_line.comments = line.comments.copy() + + append_leaves(new_line, line, LL[: string_idx - 1]) + + string_leaf = Leaf(token.STRING, LL[string_idx].value) + LL[string_idx - 1].remove() + replace_child(LL[string_idx], string_leaf) + new_line.append(string_leaf) + + append_leaves( + new_line, line, LL[string_idx + 1 : rpar_idx] + LL[rpar_idx + 1 :], + ) + + LL[rpar_idx].remove() + + yield Ok(new_line) + + +class BaseStringSplitter(StringTransformer): + """ + Abstract class for StringTransformers which transform a Line's strings by splitting + them or placing them on their own lines where necessary to avoid going over + the configured line length. + + Requirements: + * The target string value is responsible for the line going over the + line length limit. It follows that after all of black's other line + split methods have been exhausted, this line (or one of the resulting + lines after all line splits are performed) would still be over the + line_length limit unless we split this string. + AND + * The target string is NOT a "pointless" string (i.e. a string that has + no parent or siblings). + AND + * The target string is not followed by an inline comment that appears + to be a pragma. + AND + * The target string is not a multiline (i.e. triple-quote) string. + """ + + @abstractmethod + def do_splitter_match(self, line: Line) -> TMatchResult: + """ + BaseStringSplitter asks its clients to override this method instead of + `StringTransformer.do_match(...)`. + + Follows the same protocol as `StringTransformer.do_match(...)`. + + Refer to `help(StringTransformer.do_match)` for more information. + """ + + def do_match(self, line: Line) -> TMatchResult: + match_result = self.do_splitter_match(line) + if isinstance(match_result, Err): + return match_result + + string_idx = match_result.ok() + vresult = self.__validate(line, string_idx) + if isinstance(vresult, Err): + return vresult + + return match_result + + def __validate(self, line: Line, string_idx: int) -> TResult[None]: + """ + Checks that @line meets all of the requirements listed in this classes' + docstring. Refer to `help(BaseStringSplitter)` for a detailed + description of those requirements. + + Returns: + * Ok(None), if ALL of the requirements are met. + OR + * Err(CannotTransform), if ANY of the requirements are NOT met. + """ + LL = line.leaves + + string_leaf = LL[string_idx] + + max_string_length = self.__get_max_string_length(line, string_idx) + if len(string_leaf.value) <= max_string_length: + return TErr( + "The string itself is not what is causing this line to be too long." + ) + + if not string_leaf.parent or [L.type for L in string_leaf.parent.children] == [ + token.STRING, + token.NEWLINE, + ]: + return TErr( + f"This string ({string_leaf.value}) appears to be pointless (i.e. has" + " no parent)." + ) + + if id(line.leaves[string_idx]) in line.comments and contains_pragma_comment( + line.comments[id(line.leaves[string_idx])] + ): + return TErr( + "Line appears to end with an inline pragma comment. Splitting the line" + " could modify the pragma's behavior." + ) + + if has_triple_quotes(string_leaf.value): + return TErr("We cannot split multiline strings.") + + return Ok(None) + + def __get_max_string_length(self, line: Line, string_idx: int) -> int: + """ + Calculates the max string length used when attempting to determine + whether or not the target string is responsible for causing the line to + go over the line length limit. + + WARNING: This method is tightly coupled to both StringSplitter and + (especially) StringParenWrapper. There is probably a better way to + accomplish what is being done here. + + Returns: + max_string_length: such that `line.leaves[string_idx].value > + max_string_length` implies that the target string IS responsible + for causing this line to exceed the line length limit. + """ + LL = line.leaves + + is_valid_index = is_valid_index_factory(LL) + + # We use the shorthand "WMA4" in comments to abbreviate "We must + # account for". When giving examples, we use STRING to mean some/any + # valid string. + # + # Finally, we use the following convenience variables: + # + # P: The leaf that is before the target string leaf. + # N: The leaf that is after the target string leaf. + # NN: The leaf that is after N. + + # WMA4 the whitespace at the beginning of the line. + offset = line.depth * 4 + + if is_valid_index(string_idx - 1): + p_idx = string_idx - 1 + if ( + LL[string_idx - 1].type == token.LPAR + and LL[string_idx - 1].value == "" + and string_idx >= 2 + ): + # If the previous leaf is an empty LPAR placeholder, we should skip it. + p_idx -= 1 + + P = LL[p_idx] + if P.type == token.PLUS: + # WMA4 a space and a '+' character (e.g. `+ STRING`). + offset += 2 + + if P.type == token.COMMA: + # WMA4 a space, a comma, and a closing bracket [e.g. `), STRING`]. + offset += 3 + + if P.type in [token.COLON, token.EQUAL, token.NAME]: + # This conditional branch is meant to handle dictionary keys, + # variable assignments, 'return STRING' statement lines, and + # 'else STRING' ternary expression lines. + + # WMA4 a single space. + offset += 1 + + # WMA4 the lengths of any leaves that came before that space. + for leaf in LL[: p_idx + 1]: + offset += len(str(leaf)) + + if is_valid_index(string_idx + 1): + N = LL[string_idx + 1] + if N.type == token.RPAR and N.value == "" and len(LL) > string_idx + 2: + # If the next leaf is an empty RPAR placeholder, we should skip it. + N = LL[string_idx + 2] + + if N.type == token.COMMA: + # WMA4 a single comma at the end of the string (e.g `STRING,`). + offset += 1 + + if is_valid_index(string_idx + 2): + NN = LL[string_idx + 2] + + if N.type == token.DOT and NN.type == token.NAME: + # This conditional branch is meant to handle method calls invoked + # off of a string literal up to and including the LPAR character. + + # WMA4 the '.' character. + offset += 1 + + if ( + is_valid_index(string_idx + 3) + and LL[string_idx + 3].type == token.LPAR + ): + # WMA4 the left parenthesis character. + offset += 1 + + # WMA4 the length of the method's name. + offset += len(NN.value) + + has_comments = False + for comment_leaf in line.comments_after(LL[string_idx]): + if not has_comments: + has_comments = True + # WMA4 two spaces before the '#' character. + offset += 2 + + # WMA4 the length of the inline comment. + offset += len(comment_leaf.value) + + max_string_length = self.line_length - offset + return max_string_length + + +class StringSplitter(CustomSplitMapMixin, BaseStringSplitter): + """ + StringTransformer that splits "atom" strings (i.e. strings which exist on + lines by themselves). + + Requirements: + * The line consists ONLY of a single string (with the exception of a + '+' symbol which MAY exist at the start of the line), MAYBE a string + trailer, and MAYBE a trailing comma. + AND + * All of the requirements listed in BaseStringSplitter's docstring. + + Transformations: + The string mentioned in the 'Requirements' section is split into as + many substrings as necessary to adhere to the configured line length. + + In the final set of substrings, no substring should be smaller than + MIN_SUBSTR_SIZE characters. + + The string will ONLY be split on spaces (i.e. each new substring should + start with a space). + + If the string is an f-string, it will NOT be split in the middle of an + f-expression (e.g. in f"FooBar: {foo() if x else bar()}", {foo() if x + else bar()} is an f-expression). + + If the string that is being split has an associated set of custom split + records and those custom splits will NOT result in any line going over + the configured line length, those custom splits are used. Otherwise the + string is split as late as possible (from left-to-right) while still + adhering to the transformation rules listed above. + + Collaborations: + StringSplitter relies on StringMerger to construct the appropriate + CustomSplit objects and add them to the custom split map. + """ + + MIN_SUBSTR_SIZE = 6 + # Matches an "f-expression" (e.g. {var}) that might be found in an f-string. + RE_FEXPR = r""" + (? TMatchResult: + LL = line.leaves + + is_valid_index = is_valid_index_factory(LL) + + idx = 0 + + # The first leaf MAY be a '+' symbol... + if is_valid_index(idx) and LL[idx].type == token.PLUS: + idx += 1 + + # The next/first leaf MAY be an empty LPAR... + if is_valid_index(idx) and is_empty_lpar(LL[idx]): + idx += 1 + + # The next/first leaf MUST be a string... + if not is_valid_index(idx) or LL[idx].type != token.STRING: + return TErr("Line does not start with a string.") + + string_idx = idx + + # Skip the string trailer, if one exists. + string_parser = StringParser() + idx = string_parser.parse(LL, string_idx) + + # That string MAY be followed by an empty RPAR... + if is_valid_index(idx) and is_empty_rpar(LL[idx]): + idx += 1 + + # That string / empty RPAR leaf MAY be followed by a comma... + if is_valid_index(idx) and LL[idx].type == token.COMMA: + idx += 1 + + # But no more leaves are allowed... + if is_valid_index(idx): + return TErr("This line does not end with a string.") + + return Ok(string_idx) + + def do_transform(self, line: Line, string_idx: int) -> Iterator[TResult[Line]]: + LL = line.leaves + + QUOTE = LL[string_idx].value[-1] + + is_valid_index = is_valid_index_factory(LL) + insert_str_child = insert_str_child_factory(LL[string_idx]) + + prefix = get_string_prefix(LL[string_idx].value) + + # We MAY choose to drop the 'f' prefix from substrings that don't + # contain any f-expressions, but ONLY if the original f-string + # containes at least one f-expression. Otherwise, we will alter the AST + # of the program. + drop_pointless_f_prefix = ("f" in prefix) and re.search( + self.RE_FEXPR, LL[string_idx].value, re.VERBOSE + ) + + first_string_line = True + starts_with_plus = LL[0].type == token.PLUS + + def line_needs_plus() -> bool: + return first_string_line and starts_with_plus + + def maybe_append_plus(new_line: Line) -> None: + """ + Side Effects: + If @line starts with a plus and this is the first line we are + constructing, this function appends a PLUS leaf to @new_line + and replaces the old PLUS leaf in the node structure. Otherwise + this function does nothing. + """ + if line_needs_plus(): + plus_leaf = Leaf(token.PLUS, "+") + replace_child(LL[0], plus_leaf) + new_line.append(plus_leaf) + + ends_with_comma = ( + is_valid_index(string_idx + 1) and LL[string_idx + 1].type == token.COMMA + ) + + def max_last_string() -> int: + """ + Returns: + The max allowed length of the string value used for the last + line we will construct. + """ + result = self.line_length + result -= line.depth * 4 + result -= 1 if ends_with_comma else 0 + result -= 2 if line_needs_plus() else 0 + return result + + # --- Calculate Max Break Index (for string value) + # We start with the line length limit + max_break_idx = self.line_length + # The last index of a string of length N is N-1. + max_break_idx -= 1 + # Leading whitespace is not present in the string value (e.g. Leaf.value). + max_break_idx -= line.depth * 4 + if max_break_idx < 0: + yield TErr( + f"Unable to split {LL[string_idx].value} at such high of a line depth:" + f" {line.depth}" + ) + return + + # Check if StringMerger registered any custom splits. + custom_splits = self.pop_custom_splits(LL[string_idx].value) + # We use them ONLY if none of them would produce lines that exceed the + # line limit. + use_custom_breakpoints = bool( + custom_splits + and all(csplit.break_idx <= max_break_idx for csplit in custom_splits) + ) + + # Temporary storage for the remaining chunk of the string line that + # can't fit onto the line currently being constructed. + rest_value = LL[string_idx].value + + def more_splits_should_be_made() -> bool: + """ + Returns: + True iff `rest_value` (the remaining string value from the last + split), should be split again. + """ + if use_custom_breakpoints: + return len(custom_splits) > 1 + else: + return len(rest_value) > max_last_string() + + string_line_results: List[Ok[Line]] = [] + while more_splits_should_be_made(): + if use_custom_breakpoints: + # Custom User Split (manual) + csplit = custom_splits.pop(0) + break_idx = csplit.break_idx + else: + # Algorithmic Split (automatic) + max_bidx = max_break_idx - 2 if line_needs_plus() else max_break_idx + maybe_break_idx = self.__get_break_idx(rest_value, max_bidx) + if maybe_break_idx is None: + # If we are unable to algorthmically determine a good split + # and this string has custom splits registered to it, we + # fall back to using them--which means we have to start + # over from the beginning. + if custom_splits: + rest_value = LL[string_idx].value + string_line_results = [] + first_string_line = True + use_custom_breakpoints = True + continue + + # Otherwise, we stop splitting here. + break + + break_idx = maybe_break_idx + + # --- Construct `next_value` + next_value = rest_value[:break_idx] + QUOTE + if ( + # Are we allowed to try to drop a pointless 'f' prefix? + drop_pointless_f_prefix + # If we are, will we be successful? + and next_value != self.__normalize_f_string(next_value, prefix) + ): + # If the current custom split did NOT originally use a prefix, + # then `csplit.break_idx` will be off by one after removing + # the 'f' prefix. + break_idx = ( + break_idx + 1 + if use_custom_breakpoints and not csplit.has_prefix + else break_idx + ) + next_value = rest_value[:break_idx] + QUOTE + next_value = self.__normalize_f_string(next_value, prefix) + + # --- Construct `next_leaf` + next_leaf = Leaf(token.STRING, next_value) + insert_str_child(next_leaf) + self.__maybe_normalize_string_quotes(next_leaf) + + # --- Construct `next_line` + next_line = line.clone() + maybe_append_plus(next_line) + next_line.append(next_leaf) + string_line_results.append(Ok(next_line)) + + rest_value = prefix + QUOTE + rest_value[break_idx:] + first_string_line = False + + yield from string_line_results + + if drop_pointless_f_prefix: + rest_value = self.__normalize_f_string(rest_value, prefix) + + rest_leaf = Leaf(token.STRING, rest_value) + insert_str_child(rest_leaf) + + # NOTE: I could not find a test case that verifies that the following + # line is actually necessary, but it seems to be. Otherwise we risk + # not normalizing the last substring, right? + self.__maybe_normalize_string_quotes(rest_leaf) + + last_line = line.clone() + maybe_append_plus(last_line) + + # If there are any leaves to the right of the target string... + if is_valid_index(string_idx + 1): + # We use `temp_value` here to determine how long the last line + # would be if we were to append all the leaves to the right of the + # target string to the last string line. + temp_value = rest_value + for leaf in LL[string_idx + 1 :]: + temp_value += str(leaf) + if leaf.type == token.LPAR: + break + + # Try to fit them all on the same line with the last substring... + if ( + len(temp_value) <= max_last_string() + or LL[string_idx + 1].type == token.COMMA + ): + last_line.append(rest_leaf) + append_leaves(last_line, line, LL[string_idx + 1 :]) + yield Ok(last_line) + # Otherwise, place the last substring on one line and everything + # else on a line below that... + else: + last_line.append(rest_leaf) + yield Ok(last_line) + + non_string_line = line.clone() + append_leaves(non_string_line, line, LL[string_idx + 1 :]) + yield Ok(non_string_line) + # Else the target string was the last leaf... + else: + last_line.append(rest_leaf) + last_line.comments = line.comments.copy() + yield Ok(last_line) + + def __get_break_idx(self, string: str, max_break_idx: int) -> Optional[int]: + """ + This method contains the algorithm that StringSplitter uses to + determine which character to split each string at. + + Args: + @string: The substring that we are attempting to split. + @max_break_idx: The ideal break index. We will return this value if it + meets all the necessary conditions. In the likely event that it + doesn't we will try to find the closest index BELOW @max_break_idx + that does. If that fails, we will expand our search by also + considering all valid indices ABOVE @max_break_idx. + + Pre-Conditions: + * assert_is_leaf_string(@string) + * 0 <= @max_break_idx < len(@string) + + Returns: + break_idx, if an index is able to be found that meets all of the + conditions listed in the 'Transformations' section of this classes' + docstring. + OR + None, otherwise. + """ + is_valid_index = is_valid_index_factory(string) + + assert is_valid_index(max_break_idx) + assert_is_leaf_string(string) + + _fexpr_slices: Optional[List[Tuple[Index, Index]]] = None + + def fexpr_slices() -> Iterator[Tuple[Index, Index]]: + """ + Yields: + All ranges of @string which, if @string were to be split there, + would result in the splitting of an f-expression (which is NOT + allowed). + """ + nonlocal _fexpr_slices + + if _fexpr_slices is None: + _fexpr_slices = [] + for match in re.finditer(self.RE_FEXPR, string, re.VERBOSE): + _fexpr_slices.append(match.span()) + + yield from _fexpr_slices + + is_fstring = "f" in get_string_prefix(string) + + def breaks_fstring_expression(i: Index) -> bool: + """ + Returns: + True iff returning @i would result in the splitting of an + f-expression (which is NOT allowed). + """ + if not is_fstring: + return False + + for (start, end) in fexpr_slices(): + if start <= i < end: + return True + + return False + + def passes_all_checks(i: Index) -> bool: + """ + Returns: + True iff ALL of the conditions listed in the 'Transformations' + section of this classes' docstring would be be met by returning @i. + """ + is_space = string[i] == " " + is_big_enough = ( + len(string[i:]) >= self.MIN_SUBSTR_SIZE + and len(string[:i]) >= self.MIN_SUBSTR_SIZE + ) + return is_space and is_big_enough and not breaks_fstring_expression(i) + + # First, we check all indices BELOW @max_break_idx. + break_idx = max_break_idx + while is_valid_index(break_idx - 1) and not passes_all_checks(break_idx): + break_idx -= 1 + + if not passes_all_checks(break_idx): + # If that fails, we check all indices ABOVE @max_break_idx. + # + # If we are able to find a valid index here, the next line is going + # to be longer than the specified line length, but it's probably + # better than doing nothing at all. + break_idx = max_break_idx + 1 + while is_valid_index(break_idx + 1) and not passes_all_checks(break_idx): + break_idx += 1 + + if not is_valid_index(break_idx) or not passes_all_checks(break_idx): + return None + + return break_idx + + def __maybe_normalize_string_quotes(self, leaf: Leaf) -> None: + if self.normalize_strings: + normalize_string_quotes(leaf) + + def __normalize_f_string(self, string: str, prefix: str) -> str: + """ + Pre-Conditions: + * assert_is_leaf_string(@string) + + Returns: + * If @string is an f-string that contains no f-expressions, we + return a string identical to @string except that the 'f' prefix + has been stripped and all double braces (i.e. '{{' or '}}') have + been normalized (i.e. turned into '{' or '}'). + OR + * Otherwise, we return @string. + """ + assert_is_leaf_string(string) + + if "f" in prefix and not re.search(self.RE_FEXPR, string, re.VERBOSE): + new_prefix = prefix.replace("f", "") + + temp = string[len(prefix) :] + temp = re.sub(r"\{\{", "{", temp) + temp = re.sub(r"\}\}", "}", temp) + new_string = temp + + return f"{new_prefix}{new_string}" + else: + return string + + +class StringParenWrapper(CustomSplitMapMixin, BaseStringSplitter): + """ + StringTransformer that splits non-"atom" strings (i.e. strings that do not + exist on lines by themselves). + + Requirements: + All of the requirements listed in BaseStringSplitter's docstring in + addition to the requirements listed below: + + * The line is a return/yield statement, which returns/yields a string. + OR + * The line is part of a ternary expression (e.g. `x = y if cond else + z`) such that the line starts with `else `, where is + some string. + OR + * The line is an assert statement, which ends with a string. + OR + * The line is an assignment statement (e.g. `x = ` or `x += + `) such that the variable is being assigned the value of some + string. + OR + * The line is a dictionary key assignment where some valid key is being + assigned the value of some string. + + Transformations: + The chosen string is wrapped in parentheses and then split at the LPAR. + + We then have one line which ends with an LPAR and another line that + starts with the chosen string. The latter line is then split again at + the RPAR. This results in the RPAR (and possibly a trailing comma) + being placed on its own line. + + NOTE: If any leaves exist to the right of the chosen string (except + for a trailing comma, which would be placed after the RPAR), those + leaves are placed inside the parentheses. In effect, the chosen + string is not necessarily being "wrapped" by parentheses. We can, + however, count on the LPAR being placed directly before the chosen + string. + + In other words, StringParenWrapper creates "atom" strings. These + can then be split again by StringSplitter, if necessary. + + Collaborations: + In the event that a string line split by StringParenWrapper is + changed such that it no longer needs to be given its own line, + StringParenWrapper relies on StringParenStripper to clean up the + parentheses it created. + """ + + def do_splitter_match(self, line: Line) -> TMatchResult: + LL = line.leaves + + string_idx = None + string_idx = string_idx or self._return_match(LL) + string_idx = string_idx or self._else_match(LL) + string_idx = string_idx or self._assert_match(LL) + string_idx = string_idx or self._assign_match(LL) + string_idx = string_idx or self._dict_match(LL) + + if string_idx is not None: + string_value = line.leaves[string_idx].value + # If the string has no spaces... + if " " not in string_value: + # And will still violate the line length limit when split... + max_string_length = self.line_length - ((line.depth + 1) * 4) + if len(string_value) > max_string_length: + # And has no associated custom splits... + if not self.has_custom_splits(string_value): + # Then we should NOT put this string on its own line. + return TErr( + "We do not wrap long strings in parentheses when the" + " resultant line would still be over the specified line" + " length and can't be split further by StringSplitter." + ) + return Ok(string_idx) + + return TErr("This line does not contain any non-atomic strings.") + + @staticmethod + def _return_match(LL: List[Leaf]) -> Optional[int]: + """ + Returns: + string_idx such that @LL[string_idx] is equal to our target (i.e. + matched) string, if this line matches the return/yield statement + requirements listed in the 'Requirements' section of this classes' + docstring. + OR + None, otherwise. + """ + # If this line is apart of a return/yield statement and the first leaf + # contains either the "return" or "yield" keywords... + if parent_type(LL[0]) in [syms.return_stmt, syms.yield_expr] and LL[ + 0 + ].value in ["return", "yield"]: + is_valid_index = is_valid_index_factory(LL) + + idx = 2 if is_valid_index(1) and is_empty_par(LL[1]) else 1 + # The next visible leaf MUST contain a string... + if is_valid_index(idx) and LL[idx].type == token.STRING: + return idx + + return None + + @staticmethod + def _else_match(LL: List[Leaf]) -> Optional[int]: + """ + Returns: + string_idx such that @LL[string_idx] is equal to our target (i.e. + matched) string, if this line matches the ternary expression + requirements listed in the 'Requirements' section of this classes' + docstring. + OR + None, otherwise. + """ + # If this line is apart of a ternary expression and the first leaf + # contains the "else" keyword... + if ( + parent_type(LL[0]) == syms.test + and LL[0].type == token.NAME + and LL[0].value == "else" + ): + is_valid_index = is_valid_index_factory(LL) + + idx = 2 if is_valid_index(1) and is_empty_par(LL[1]) else 1 + # The next visible leaf MUST contain a string... + if is_valid_index(idx) and LL[idx].type == token.STRING: + return idx + + return None + + @staticmethod + def _assert_match(LL: List[Leaf]) -> Optional[int]: + """ + Returns: + string_idx such that @LL[string_idx] is equal to our target (i.e. + matched) string, if this line matches the assert statement + requirements listed in the 'Requirements' section of this classes' + docstring. + OR + None, otherwise. + """ + # If this line is apart of an assert statement and the first leaf + # contains the "assert" keyword... + if parent_type(LL[0]) == syms.assert_stmt and LL[0].value == "assert": + is_valid_index = is_valid_index_factory(LL) + + for (i, leaf) in enumerate(LL): + # We MUST find a comma... + if leaf.type == token.COMMA: + idx = i + 2 if is_empty_par(LL[i + 1]) else i + 1 + + # That comma MUST be followed by a string... + if is_valid_index(idx) and LL[idx].type == token.STRING: + string_idx = idx + + # Skip the string trailer, if one exists. + string_parser = StringParser() + idx = string_parser.parse(LL, string_idx) + + # But no more leaves are allowed... + if not is_valid_index(idx): + return string_idx + + return None + + @staticmethod + def _assign_match(LL: List[Leaf]) -> Optional[int]: + """ + Returns: + string_idx such that @LL[string_idx] is equal to our target (i.e. + matched) string, if this line matches the assignment statement + requirements listed in the 'Requirements' section of this classes' + docstring. + OR + None, otherwise. + """ + # If this line is apart of an expression statement or is a function + # argument AND the first leaf contains a variable name... + if ( + parent_type(LL[0]) in [syms.expr_stmt, syms.argument, syms.power] + and LL[0].type == token.NAME + ): + is_valid_index = is_valid_index_factory(LL) + + for (i, leaf) in enumerate(LL): + # We MUST find either an '=' or '+=' symbol... + if leaf.type in [token.EQUAL, token.PLUSEQUAL]: + idx = i + 2 if is_empty_par(LL[i + 1]) else i + 1 + + # That symbol MUST be followed by a string... + if is_valid_index(idx) and LL[idx].type == token.STRING: + string_idx = idx + + # Skip the string trailer, if one exists. + string_parser = StringParser() + idx = string_parser.parse(LL, string_idx) + + # The next leaf MAY be a comma iff this line is apart + # of a function argument... + if ( + parent_type(LL[0]) == syms.argument + and is_valid_index(idx) + and LL[idx].type == token.COMMA + ): + idx += 1 + + # But no more leaves are allowed... + if not is_valid_index(idx): + return string_idx + + return None + + @staticmethod + def _dict_match(LL: List[Leaf]) -> Optional[int]: + """ + Returns: + string_idx such that @LL[string_idx] is equal to our target (i.e. + matched) string, if this line matches the dictionary key assignment + statement requirements listed in the 'Requirements' section of this + classes' docstring. + OR + None, otherwise. + """ + # If this line is apart of a dictionary key assignment... + if syms.dictsetmaker in [parent_type(LL[0]), parent_type(LL[0].parent)]: + is_valid_index = is_valid_index_factory(LL) + + for (i, leaf) in enumerate(LL): + # We MUST find a colon... + if leaf.type == token.COLON: + idx = i + 2 if is_empty_par(LL[i + 1]) else i + 1 + + # That colon MUST be followed by a string... + if is_valid_index(idx) and LL[idx].type == token.STRING: + string_idx = idx + + # Skip the string trailer, if one exists. + string_parser = StringParser() + idx = string_parser.parse(LL, string_idx) + + # That string MAY be followed by a comma... + if is_valid_index(idx) and LL[idx].type == token.COMMA: + idx += 1 + + # But no more leaves are allowed... + if not is_valid_index(idx): + return string_idx + + return None + + def do_transform(self, line: Line, string_idx: int) -> Iterator[TResult[Line]]: + LL = line.leaves + + is_valid_index = is_valid_index_factory(LL) + insert_str_child = insert_str_child_factory(LL[string_idx]) + + comma_idx = len(LL) - 1 + ends_with_comma = False + if LL[comma_idx].type == token.COMMA: + ends_with_comma = True + + leaves_to_steal_comments_from = [LL[string_idx]] + if ends_with_comma: + leaves_to_steal_comments_from.append(LL[comma_idx]) + + # --- First Line + first_line = line.clone() + left_leaves = LL[:string_idx] + + # We have to remember to account for (possibly invisible) LPAR and RPAR + # leaves that already wrapped the target string. If these leaves do + # exist, we will replace them with our own LPAR and RPAR leaves. + old_parens_exist = False + if left_leaves and left_leaves[-1].type == token.LPAR: + old_parens_exist = True + leaves_to_steal_comments_from.append(left_leaves[-1]) + left_leaves.pop() + + append_leaves(first_line, line, left_leaves) + + lpar_leaf = Leaf(token.LPAR, "(") + if old_parens_exist: + replace_child(LL[string_idx - 1], lpar_leaf) + else: + insert_str_child(lpar_leaf) + first_line.append(lpar_leaf) + + # We throw inline comments that were originally to the right of the + # target string to the top line. They will now be shown to the right of + # the LPAR. + for leaf in leaves_to_steal_comments_from: + for comment_leaf in line.comments_after(leaf): + first_line.append(comment_leaf, preformatted=True) + + yield Ok(first_line) + + # --- Middle (String) Line + # We only need to yield one (possibly too long) string line, since the + # `StringSplitter` will break it down further if necessary. + string_value = LL[string_idx].value + string_line = Line( + depth=line.depth + 1, + inside_brackets=True, + should_explode=line.should_explode, + ) + string_leaf = Leaf(token.STRING, string_value) + insert_str_child(string_leaf) + string_line.append(string_leaf) + + old_rpar_leaf = None + if is_valid_index(string_idx + 1): + right_leaves = LL[string_idx + 1 :] + if ends_with_comma: + right_leaves.pop() + + if old_parens_exist: + assert ( + right_leaves and right_leaves[-1].type == token.RPAR + ), "Apparently, old parentheses do NOT exist?!" + old_rpar_leaf = right_leaves.pop() + + append_leaves(string_line, line, right_leaves) + + yield Ok(string_line) + + # --- Last Line + last_line = line.clone() + last_line.bracket_tracker = first_line.bracket_tracker + + new_rpar_leaf = Leaf(token.RPAR, ")") + if old_rpar_leaf is not None: + replace_child(old_rpar_leaf, new_rpar_leaf) + else: + insert_str_child(new_rpar_leaf) + last_line.append(new_rpar_leaf) + + # If the target string ended with a comma, we place this comma to the + # right of the RPAR on the last line. + if ends_with_comma: + comma_leaf = Leaf(token.COMMA, ",") + replace_child(LL[comma_idx], comma_leaf) + last_line.append(comma_leaf) + + yield Ok(last_line) + + +class StringParser: + """ + A state machine that aids in parsing a string's "trailer", which can be + either non-existant, an old-style formatting sequence (e.g. `% varX` or `% + (varX, varY)`), or a method-call / attribute access (e.g. `.format(varX, + varY)`). + + NOTE: A new StringParser object MUST be instantiated for each string + trailer we need to parse. + + Examples: + We shall assume that `line` equals the `Line` object that corresponds + to the following line of python code: + ``` + x = "Some {}.".format("String") + some_other_string + ``` + + Furthermore, we will assume that `string_idx` is some index such that: + ``` + assert line.leaves[string_idx].value == "Some {}." + ``` + + The following code snippet then holds: + ``` + string_parser = StringParser() + idx = string_parser.parse(line.leaves, string_idx) + assert line.leaves[idx].type == token.PLUS + ``` + """ + + DEFAULT_TOKEN = -1 + + # String Parser States + START = 1 + DOT = 2 + NAME = 3 + PERCENT = 4 + SINGLE_FMT_ARG = 5 + LPAR = 6 + RPAR = 7 + DONE = 8 + + # Lookup Table for Next State + _goto: Dict[Tuple[ParserState, NodeType], ParserState] = { + # A string trailer may start with '.' OR '%'. + (START, token.DOT): DOT, + (START, token.PERCENT): PERCENT, + (START, DEFAULT_TOKEN): DONE, + # A '.' MUST be followed by an attribute or method name. + (DOT, token.NAME): NAME, + # A method name MUST be followed by an '(', whereas an attribute name + # is the last symbol in the string trailer. + (NAME, token.LPAR): LPAR, + (NAME, DEFAULT_TOKEN): DONE, + # A '%' symbol can be followed by an '(' or a single argument (e.g. a + # string or variable name). + (PERCENT, token.LPAR): LPAR, + (PERCENT, DEFAULT_TOKEN): SINGLE_FMT_ARG, + # If a '%' symbol is followed by a single argument, that argument is + # the last leaf in the string trailer. + (SINGLE_FMT_ARG, DEFAULT_TOKEN): DONE, + # If present, a ')' symbol is the last symbol in a string trailer. + # (NOTE: LPARS and nested RPARS are not included in this lookup table, + # since they are treated as a special case by the parsing logic in this + # classes' implementation.) + (RPAR, DEFAULT_TOKEN): DONE, + } + + def __init__(self) -> None: + self._state = self.START + self._unmatched_lpars = 0 + + def parse(self, leaves: List[Leaf], string_idx: int) -> int: + """ + Pre-conditions: + * @leaves[@string_idx].type == token.STRING + + Returns: + The index directly after the last leaf which is apart of the string + trailer, if a "trailer" exists. + OR + @string_idx + 1, if no string "trailer" exists. + """ + assert leaves[string_idx].type == token.STRING + + idx = string_idx + 1 + while idx < len(leaves) and self._next_state(leaves[idx]): + idx += 1 + return idx + + def _next_state(self, leaf: Leaf) -> bool: + """ + Pre-conditions: + * On the first call to this function, @leaf MUST be the leaf that + was directly after the string leaf in question (e.g. if our target + string is `line.leaves[i]` then the first call to this method must + be `line.leaves[i + 1]`). + * On the next call to this function, the leaf paramater passed in + MUST be the leaf directly following @leaf. + + Returns: + True iff @leaf is apart of the string's trailer. + """ + # We ignore empty LPAR or RPAR leaves. + if is_empty_par(leaf): + return True + + next_token = leaf.type + if next_token == token.LPAR: + self._unmatched_lpars += 1 + + current_state = self._state + + # The LPAR parser state is a special case. We will return True until we + # find the matching RPAR token. + if current_state == self.LPAR: + if next_token == token.RPAR: + self._unmatched_lpars -= 1 + if self._unmatched_lpars == 0: + self._state = self.RPAR + # Otherwise, we use a lookup table to determine the next state. + else: + # If the lookup table matches the current state to the next + # token, we use the lookup table. + if (current_state, next_token) in self._goto: + self._state = self._goto[current_state, next_token] + else: + # Otherwise, we check if a the current state was assigned a + # default. + if (current_state, self.DEFAULT_TOKEN) in self._goto: + self._state = self._goto[current_state, self.DEFAULT_TOKEN] + # If no default has been assigned, then this parser has a logic + # error. + else: + raise RuntimeError(f"{self.__class__.__name__} LOGIC ERROR!") + + if self._state == self.DONE: + return False + + return True + + +def TErr(err_msg: str) -> Err[CannotTransform]: + """(T)ransform Err + + Convenience function used when working with the TResult type. + """ + cant_transform = CannotTransform(err_msg) + return Err(cant_transform) + + +def contains_pragma_comment(comment_list: List[Leaf]) -> bool: + """ + Returns: + True iff one of the comments in @comment_list is a pragma used by one + of the more common static analysis tools for python (e.g. mypy, flake8, + pylint). + """ + for comment in comment_list: + if comment.value.startswith(("# type:", "# noqa", "# pylint:")): + return True + + return False + + +def insert_str_child_factory(string_leaf: Leaf) -> Callable[[LN], None]: + """ + Factory for a convenience function that is used to orphan @string_leaf + and then insert multiple new leaves into the same part of the node + structure that @string_leaf had originally occupied. + + Examples: + Let `string_leaf = Leaf(token.STRING, '"foo"')` and `N = + string_leaf.parent`. Assume the node `N` has the following + original structure: + + Node( + expr_stmt, [ + Leaf(NAME, 'x'), + Leaf(EQUAL, '='), + Leaf(STRING, '"foo"'), + ] + ) + + We then run the code snippet shown below. + ``` + insert_str_child = insert_str_child_factory(string_leaf) + + lpar = Leaf(token.LPAR, '(') + insert_str_child(lpar) + + bar = Leaf(token.STRING, '"bar"') + insert_str_child(bar) + + rpar = Leaf(token.RPAR, ')') + insert_str_child(rpar) + ``` + + After which point, it follows that `string_leaf.parent is None` and + the node `N` now has the following structure: + + Node( + expr_stmt, [ + Leaf(NAME, 'x'), + Leaf(EQUAL, '='), + Leaf(LPAR, '('), + Leaf(STRING, '"bar"'), + Leaf(RPAR, ')'), + ] + ) + """ + string_parent = string_leaf.parent + string_child_idx = string_leaf.remove() + + def insert_str_child(child: LN) -> None: + nonlocal string_child_idx + + assert string_parent is not None + assert string_child_idx is not None + + string_parent.insert_child(string_child_idx, child) + string_child_idx += 1 + + return insert_str_child + + +def has_triple_quotes(string: str) -> bool: + """ + Returns: + True iff @string starts with three quotation characters. + """ + raw_string = string.lstrip(STRING_PREFIX_CHARS) + return raw_string[:3] in {'"""', "'''"} + + +def parent_type(node: Optional[LN]) -> Optional[NodeType]: + """ + Returns: + @node.parent.type, if @node is not None and has a parent. + OR + None, otherwise. + """ + if node is None or node.parent is None: + return None + + return node.parent.type + + +def is_empty_par(leaf: Leaf) -> bool: + return is_empty_lpar(leaf) or is_empty_rpar(leaf) + + +def is_empty_lpar(leaf: Leaf) -> bool: + return leaf.type == token.LPAR and leaf.value == "" + + +def is_empty_rpar(leaf: Leaf) -> bool: + return leaf.type == token.RPAR and leaf.value == "" + + +def is_valid_index_factory(seq: Sequence[Any]) -> Callable[[int], bool]: + """ + Examples: + ``` + my_list = [1, 2, 3] + + is_valid_index = is_valid_index_factory(my_list) + + assert is_valid_index(0) + assert is_valid_index(2) + + assert not is_valid_index(3) + assert not is_valid_index(-1) + ``` + """ + + def is_valid_index(idx: int) -> bool: + """ + Returns: + True iff @idx is positive AND seq[@idx] does NOT raise an + IndexError. + """ + return 0 <= idx < len(seq) + + return is_valid_index + + +def line_to_string(line: Line) -> str: + """Returns the string representation of @line. + + WARNING: This is known to be computationally expensive. + """ + return str(line).strip("\n") + + +def append_leaves(new_line: Line, old_line: Line, leaves: List[Leaf]) -> None: + """ + Append leaves (taken from @old_line) to @new_line, making sure to fix the + underlying Node structure where appropriate. + + All of the leaves in @leaves are duplicated. The duplicates are then + appended to @new_line and used to replace their originals in the underlying + Node structure. Any comments attatched to the old leaves are reattached to + the new leaves. + + Pre-conditions: + set(@leaves) is a subset of set(@old_line.leaves). + """ + for old_leaf in leaves: + assert old_leaf in old_line.leaves + + new_leaf = Leaf(old_leaf.type, old_leaf.value) + replace_child(old_leaf, new_leaf) + new_line.append(new_leaf) + + for comment_leaf in old_line.comments_after(old_leaf): + new_line.append(comment_leaf, preformatted=True) + + +def replace_child(old_child: LN, new_child: LN) -> None: + """ + Side Effects: + * If @old_child.parent is set, replace @old_child with @new_child in + @old_child's underlying Node structure. + OR + * Otherwise, this function does nothing. + """ + parent = old_child.parent + if not parent: + return + + child_idx = old_child.remove() + if child_idx is not None: + parent.insert_child(child_idx, new_child) + + +def get_string_prefix(string: str) -> str: + """ + Pre-conditions: + * assert_is_leaf_string(@string) + + Returns: + @string's prefix (e.g. '', 'r', 'f', or 'rf'). + """ + assert_is_leaf_string(string) + + prefix = "" + prefix_idx = 0 + while string[prefix_idx] in STRING_PREFIX_CHARS: + prefix += string[prefix_idx].lower() + prefix_idx += 1 + + return prefix + + +def assert_is_leaf_string(string: str) -> None: + """ + Checks the pre-condition that @string has the format that you would expect + of `leaf.value` where `leaf` is some Leaf such that `leaf.type == + token.STRING`. A more precise description of the pre-conditions that are + checked are listed below. + + Pre-conditions: + * @string starts with either ', ", ', or " where + `set()` is some subset of `set(STRING_PREFIX_CHARS)`. + * @string ends with a quote character (' or "). + + Raises: + AssertionError(...) if the pre-conditions listed above are not + satisfied. + """ + dquote_idx = string.find('"') + squote_idx = string.find("'") + if -1 in [dquote_idx, squote_idx]: + quote_idx = max(dquote_idx, squote_idx) + else: + quote_idx = min(squote_idx, dquote_idx) + + assert ( + 0 <= quote_idx < len(string) - 1 + ), f"{string!r} is missing a starting quote character (' or \")." + assert string[-1] in ( + "'", + '"', + ), f"{string!r} is missing an ending quote character (' or \")." + assert set(string[:quote_idx]).issubset( + set(STRING_PREFIX_CHARS) + ), f"{set(string[:quote_idx])} is NOT a subset of {set(STRING_PREFIX_CHARS)}." + + +def left_hand_split(line: Line, _features: Collection[Feature] = ()) -> Iterator[Line]: + """Split line into many lines, starting with the first matching bracket pair. + + Note: this usually looks weird, only use this for function definitions. + Prefer RHS otherwise. This is why this function is not symmetrical with + :func:`right_hand_split` which also handles optional parentheses. + """ + tail_leaves: List[Leaf] = [] + body_leaves: List[Leaf] = [] + head_leaves: List[Leaf] = [] + current_leaves = head_leaves + matching_bracket: Optional[Leaf] = None + for leaf in line.leaves: + if ( + current_leaves is body_leaves + and leaf.type in CLOSING_BRACKETS + and leaf.opening_bracket is matching_bracket + ): + current_leaves = tail_leaves if body_leaves else head_leaves current_leaves.append(leaf) if current_leaves is head_leaves: if leaf.type in OPENING_BRACKETS: matching_bracket = leaf current_leaves = body_leaves - # Since body is a new indent level, remove spurious leading whitespace. - 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 leaf in leaves: - result.append(leaf, preformatted=True) - for comment_after in line.comments_after(leaf): - result.append(comment_after, preformatted=True) + if not matching_bracket: + raise CannotSplit("No brackets found") + + head = bracket_split_build_line(head_leaves, line, matching_bracket) + body = bracket_split_build_line(body_leaves, line, matching_bracket, is_body=True) + tail = bracket_split_build_line(tail_leaves, line, matching_bracket) bracket_split_succeeded_or_raise(head, body, tail) for result in (head, body, tail): if result: @@ -1905,21 +4661,25 @@ def left_hand_split(line: Line, py36: bool = False) -> Iterator[Line]: def right_hand_split( - line: Line, py36: bool = False, omit: Collection[LeafID] = () + line: Line, + line_length: int, + features: Collection[Feature] = (), + omit: Collection[LeafID] = (), ) -> Iterator[Line]: """Split line into many lines, starting with the last matching bracket pair. If the split was by optional parentheses, attempt splitting without them, too. + `omit` is a collection of closing bracket IDs that shouldn't be considered for + this split. + + Note: running this function modifies `bracket_depth` on the leaves of `line`. """ - head = Line(depth=line.depth) - body = Line(depth=line.depth + 1, inside_brackets=True) - tail = Line(depth=line.depth) tail_leaves: List[Leaf] = [] body_leaves: List[Leaf] = [] head_leaves: List[Leaf] = [] current_leaves = tail_leaves - opening_bracket = None - closing_bracket = None + opening_bracket: Optional[Leaf] = None + closing_bracket: Optional[Leaf] = None for leaf in reversed(line.leaves): if current_leaves is body_leaves: if leaf is opening_bracket: @@ -1930,44 +4690,57 @@ def right_hand_split( opening_bracket = leaf.opening_bracket closing_bracket = leaf current_leaves = body_leaves + if not (opening_bracket and closing_bracket and head_leaves): + # If there is no opening or closing_bracket that means the split failed and + # all content is in the tail. Otherwise, if `head_leaves` are empty, it means + # the matching `opening_bracket` wasn't available on `line` anymore. + raise CannotSplit("No brackets found") + tail_leaves.reverse() body_leaves.reverse() head_leaves.reverse() - # 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 leaf in leaves: - result.append(leaf, preformatted=True) - for comment_after in line.comments_after(leaf): - result.append(comment_after, preformatted=True) + head = bracket_split_build_line(head_leaves, line, opening_bracket) + body = bracket_split_build_line(body_leaves, line, opening_bracket, is_body=True) + tail = bracket_split_build_line(tail_leaves, line, opening_bracket) bracket_split_succeeded_or_raise(head, body, tail) - assert opening_bracket and closing_bracket 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 delimiters or standalone comments in the body - and not body.bracket_tracker.delimiters - 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} try: - yield from right_hand_split(line, py36=py36, omit=omit) + yield from right_hand_split(line, line_length, features=features, omit=omit) return + except CannotSplit: - pass + 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) @@ -1997,20 +4770,62 @@ def bracket_split_succeeded_or_raise(head: Line, body: Line, tail: Line) -> None elif tail_len < 3: raise CannotSplit( - f"Splitting brackets on an empty body to save " - f"{tail_len} characters is not worth it" + f"Splitting brackets on an empty body to save {tail_len} characters is" + " not worth it" ) -def dont_increase_indentation(split_func: SplitFunc) -> SplitFunc: +def bracket_split_build_line( + leaves: List[Leaf], original: Line, opening_bracket: Leaf, *, is_body: bool = False +) -> Line: + """Return a new line with given `leaves` and respective comments from `original`. + + If `is_body` is True, the result line is one-indented inside brackets and as such + has its first leaf's prefix normalized and a trailing comma added when expected. + """ + result = Line(depth=original.depth) + if is_body: + result.inside_brackets = True + result.depth += 1 + if leaves: + # Since body is a new indent level, remove spurious leading whitespace. + normalize_prefix(leaves[0], inside_brackets=True) + # Ensure a trailing comma for imports and standalone function arguments, but + # be careful not to add one after any comments or within type annotations. + no_commas = ( + original.is_def + and opening_bracket.value == "(" + 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 + + if leaves[i].type != token.COMMA: + leaves.insert(i + 1, Leaf(token.COMMA, ",")) + break + + # Populate the line + for leaf in leaves: + result.append(leaf, preformatted=True) + for comment_after in original.comments_after(leaf): + result.append(comment_after, preformatted=True) + if is_body: + result.should_explode = should_explode(result, opening_bracket) + return result + + +def dont_increase_indentation(split_func: Transformer) -> Transformer: """Normalize prefix of the first leaf in every line returned by `split_func`. This is a decorator over relevant split functions. """ @wraps(split_func) - def split_wrapper(line: Line, py36: bool = False) -> Iterator[Line]: - for l in split_func(line, py36): + def split_wrapper(line: Line, features: Collection[Feature] = ()) -> Iterator[Line]: + for l in split_func(line, features): normalize_prefix(l.leaves[0], inside_brackets=True) yield l @@ -2018,25 +4833,27 @@ def dont_increase_indentation(split_func: SplitFunc) -> SplitFunc: @dont_increase_indentation -def delimiter_split(line: Line, py36: bool = False) -> Iterator[Line]: +def delimiter_split(line: Line, features: Collection[Feature] = ()) -> Iterator[Line]: """Split according to delimiters of the highest priority. - If `py36` is True, the split will add trailing commas also in function - signatures that contain `*` and `**`. + If the appropriate Features are given, the split will add trailing commas + also in function signatures and calls that contain `*` and `**`. """ try: last_leaf = line.leaves[-1] except IndexError: raise CannotSplit("Line empty") - delimiters = line.bracket_tracker.delimiters + bt = line.bracket_tracker try: - delimiter_priority = line.bracket_tracker.max_delimiter_priority( - exclude={id(last_leaf)} - ) + delimiter_priority = bt.max_delimiter_priority(exclude={id(last_leaf)}) except ValueError: raise CannotSplit("No delimiters found") + if delimiter_priority == DOT_PRIORITY: + if bt.delimiter_count_with_priority(delimiter_priority) == 1: + raise CannotSplit("Splitting a single attribute from its owner looks wrong") + current_line = Line(depth=line.depth, inside_brackets=line.inside_brackets) lowest_depth = sys.maxsize trailing_comma_safe = True @@ -2046,7 +4863,7 @@ def delimiter_split(line: Line, py36: bool = False) -> Iterator[Line]: nonlocal current_line try: current_line.append_safe(leaf, preformatted=True) - except ValueError as ve: + except ValueError: yield current_line current_line = Line(depth=line.depth, inside_brackets=line.inside_brackets) @@ -2059,12 +4876,17 @@ def delimiter_split(line: Line, py36: bool = False) -> Iterator[Line]: yield from append_to_line(comment_after) lowest_depth = min(lowest_depth, leaf.bracket_depth) - if ( - leaf.bracket_depth == lowest_depth - and is_vararg(leaf, within=VARARGS_PARENTS) - ): - trailing_comma_safe = trailing_comma_safe and py36 - leaf_priority = delimiters.get(id(leaf)) + if leaf.bracket_depth == lowest_depth: + if is_vararg(leaf, within={syms.typedargslist}): + trailing_comma_safe = ( + trailing_comma_safe and Feature.TRAILING_COMMA_IN_DEF in features + ) + elif is_vararg(leaf, within={syms.arglist, syms.argument}): + trailing_comma_safe = ( + trailing_comma_safe and Feature.TRAILING_COMMA_IN_CALL in features + ) + + leaf_priority = bt.delimiters.get(id(leaf)) if leaf_priority == delimiter_priority: yield current_line @@ -2081,7 +4903,9 @@ def delimiter_split(line: Line, py36: bool = False) -> Iterator[Line]: @dont_increase_indentation -def standalone_comment_split(line: Line, py36: bool = False) -> Iterator[Line]: +def standalone_comment_split( + line: Line, features: Collection[Feature] = () +) -> Iterator[Line]: """Split standalone comments from the rest of the line.""" if not line.contains_standalone_comments(0): raise CannotSplit("Line does not have any standalone comments") @@ -2093,7 +4917,7 @@ def standalone_comment_split(line: Line, py36: bool = False) -> Iterator[Line]: nonlocal current_line try: current_line.append_safe(leaf, preformatted=True) - except ValueError as ve: + except ValueError: yield current_line current_line = Line(depth=line.depth, inside_brackets=line.inside_brackets) @@ -2109,26 +4933,6 @@ def standalone_comment_split(line: Line, py36: bool = False) -> Iterator[Line]: yield current_line -def explode_split( - line: Line, py36: bool = False, omit: Collection[LeafID] = () -) -> Iterator[Line]: - """Split by rightmost bracket and immediately split contents by a delimiter.""" - new_lines = list(right_hand_split(line, py36, omit)) - if len(new_lines) != 3: - yield from new_lines - return - - yield new_lines[0] - - try: - yield from delimiter_split(new_lines[1], py36) - - except CannotSplit: - yield new_lines[1] - - yield new_lines[2] - - def is_import(leaf: Leaf) -> bool: """Return True if the given leaf starts an import statement.""" p = leaf.parent @@ -2143,6 +4947,14 @@ def is_import(leaf: Leaf) -> bool: ) +def is_type_comment(leaf: Leaf, suffix: str = "") -> bool: + """Return True if the given leaf is a special comment. + Only returns true for type comments for now.""" + t = leaf.type + v = leaf.value + return t in {token.COMMENT, STANDALONE_COMMENT} and v.startswith("# type:" + suffix) + + def normalize_prefix(leaf: Leaf, *, inside_brackets: bool) -> None: """Leave existing extra newlines if not `inside_brackets`. Remove everything else. @@ -2161,6 +4973,22 @@ def normalize_prefix(leaf: Leaf, *, inside_brackets: bool) -> None: leaf.prefix = "" +def normalize_string_prefix(leaf: Leaf, remove_u_prefix: bool = False) -> None: + """Make all string prefixes lowercase. + + If remove_u_prefix is given, also removes any u prefix from the string. + + Note: Mutates its argument. + """ + match = re.match(r"^([" + STRING_PREFIX_CHARS + r"]*)(.*)$", leaf.value, re.DOTALL) + assert match is not None, f"failed to match string {leaf.value!r}" + orig_prefix = match.group(1) + new_prefix = orig_prefix.replace("F", "f").replace("B", "b").replace("U", "u") + if remove_u_prefix: + new_prefix = new_prefix.replace("u", "") + leaf.value = f"{new_prefix}{match.group(2)}" + + def normalize_string_quotes(leaf: Leaf) -> None: """Prefer double quotes but only if it doesn't cause more escaping. @@ -2169,7 +4997,7 @@ def normalize_string_quotes(leaf: Leaf) -> None: Note: Mutates its argument. """ - value = leaf.value.lstrip("furbFURB") + value = leaf.value.lstrip(STRING_PREFIX_CHARS) if value[:3] == '"""': return @@ -2188,8 +5016,8 @@ 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): @@ -2200,15 +5028,30 @@ 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""" + (?:[^{]|^)\{ # 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 + return + + if new_quote == '"""' and new_body[-1:] == '"': # edge case: new_body = new_body[:-1] + '\\"' orig_escape_count = body.count("\\") @@ -2222,43 +5065,238 @@ def normalize_string_quotes(leaf: Leaf) -> None: leaf.value = f"{prefix}{new_quote}{new_body}{new_quote}" +def normalize_numeric_literal(leaf: Leaf) -> None: + """Normalizes numeric (float, int, and complex) literals. + + All letters used in the representation are normalized to lowercase (except + in Python 2 long literals). + """ + text = leaf.value.lower() + if text.startswith(("0o", "0b")): + # Leave octal and binary literals alone. + pass + elif text.startswith("0x"): + # Change hex literals to upper case. + before, after = text[:2], text[2:] + text = f"{before}{after.upper()}" + 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) + text = f"{before}e{sign}{after}" + elif text.endswith(("j", "l")): + number = text[:-1] + suffix = text[-1] + # Capitalize in "2L" because "l" looks too similar to "1". + if suffix == "l": + suffix = "L" + text = f"{format_float_or_int_string(number)}{suffix}" + else: + text = format_float_or_int_string(text) + leaf.value = text + + +def format_float_or_int_string(text: str) -> str: + """Formats a float string like "1.0".""" + if "." not in text: + return text + + before, after = text.split(".") + return f"{before or 0}.{after or 0}" + + 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 - 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: - maybe_make_parens_invisible_in_atom(child) - 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])) + Standardizes on visible parentheses for single-element tuples, and keeps + existing visible parentheses for other tuples and generator expressions. + """ + 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)): + # Fixes a bug where invisible parens are not properly stripped from + # assignment statements that contain type annotations. + if isinstance(child, Node) and child.type == syms.annassign: + normalize_invisible_parens(child, parens_after=parens_after) + + # 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 is_walrus_assignment(child): + continue + + if child.type == syms.atom: + if maybe_make_parens_invisible_in_atom(child, parent=node): + wrap_in_parentheses(node, child, visible=False) + elif is_one_tuple(child): + wrap_in_parentheses(node, child, visible=True) + elif node.type == syms.import_from: + # "import from" nodes store parentheses directly as part of + # the statement + if child.type == token.LPAR: + # make parentheses invisible + child.value = "" # type: ignore + node.children[-1].value = "" # type: ignore + elif child.type != token.STAR: + # insert invisible parentheses + node.insert_child(index, Leaf(token.LPAR, "")) + node.append_child(Leaf(token.RPAR, "")) + break + + elif not (isinstance(child, Leaf) and is_multiline_string(child)): + wrap_in_parentheses(node, child, visible=False) + + 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: Optional[int] = 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: + if is_fmt_on(container): + return + + # fix for fmt: on in children + if contains_fmt_on_at_column(container, leaf.column): + for child in container.children: + if contains_fmt_on_at_column(child, leaf.column): + return + yield child + else: + yield container + container = container.next_sibling + + +def is_fmt_on(container: LN) -> bool: + """Determine whether formatting is switched on within a container. + Determined by whether the last `# fmt:` comment is `on` or `off`. + """ + fmt_on = False + for comment in list_comments(container.prefix, is_endmarker=False): + if comment.value in FMT_ON: + fmt_on = True + elif comment.value in FMT_OFF: + fmt_on = False + return fmt_on + + +def contains_fmt_on_at_column(container: LN, column: int) -> bool: + """Determine if children at a given column have formatting switched on.""" + for child in container.children: + if ( + isinstance(child, Node) + and first_leaf_column(child) == column + or isinstance(child, Leaf) + and child.column == column + ): + if is_fmt_on(child): + return True + + return False + + +def first_leaf_column(node: Node) -> Optional[int]: + """Returns the column of the first leaf child of a node.""" + for child in node.children: + if isinstance(child, Leaf): + return child.column + return None + - check_lpar = isinstance(child, Leaf) and child.value in parens_after +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. + Additionally, remove repeated, adjacent invisible parens from the atom `node` + as they are redundant. + Returns whether the node should itself be wrapped in invisible parentheses. -def maybe_make_parens_invisible_in_atom(node: LN) -> bool: - """If it's safe, make the parens in the atom `node` invisible, recusively.""" + """ if ( 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 @@ -2266,14 +5304,38 @@ def maybe_make_parens_invisible_in_atom(node: LN) -> bool: first = node.children[0] last = node.children[-1] if first.type == token.LPAR and last.type == token.RPAR: + middle = node.children[1] # make parentheses invisible first.value = "" # type: ignore last.value = "" # type: ignore - if len(node.children) > 1: - maybe_make_parens_invisible_in_atom(node.children[1]) - return True + maybe_make_parens_invisible_in_atom(middle, parent=parent) - return False + if is_atom_with_invisible_parens(middle): + # Strip the invisible parens from `middle` by replacing + # it with the child in-between the invisible parens + middle.replace(middle.children[1]) + + return False + + return True + + +def is_atom_with_invisible_parens(node: LN) -> bool: + """Given a `LN`, determines whether it's an atom `node` with invisible + parens. Useful in dedupe-ing and normalizing parens. + """ + if isinstance(node, Leaf) or node.type != syms.atom: + return False + + first, last = node.children[0], node.children[-1] + return ( + isinstance(first, Leaf) + and first.type == token.LPAR + and first.value == "" + and isinstance(last, Leaf) + and last.type == token.RPAR + and last.value == "" + ) def is_empty_tuple(node: LN) -> bool: @@ -2286,18 +5348,43 @@ def is_empty_tuple(node: LN) -> bool: ) +def unwrap_singleton_parenthesis(node: LN) -> Optional[LN]: + """Returns `wrapped` if `node` is of the shape ( wrapped ). + + Parenthesis can be optional. Returns None otherwise""" + if len(node.children) != 3: + return None + + lpar, wrapped, rpar = node.children + if not (lpar.type == token.LPAR and rpar.type == token.RPAR): + return None + + return wrapped + + +def wrap_in_parentheses(parent: Node, child: LN, *, visible: bool = True) -> None: + """Wrap `child` in parentheses. + + This replaces `child` with an atom holding the parentheses and the old + child. That requires moving the prefix. + + If `visible` is False, the leaves will be valueless (and thus invisible). + """ + lpar = Leaf(token.LPAR, "(" if visible else "") + rpar = Leaf(token.RPAR, ")" if visible else "") + prefix = child.prefix + child.prefix = "" + index = child.remove() or 0 + new_child = Node(syms.atom, [lpar, child, rpar]) + new_child.prefix = prefix + parent.insert_child(index, new_child) + + 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 - ): + gexp = unwrap_singleton_parenthesis(node) + if gexp is None or gexp.type != syms.testlist_gexp: return False return len(gexp.children) == 2 and gexp.children[1].type == token.COMMA @@ -2309,6 +5396,12 @@ def is_one_tuple(node: LN) -> bool: ) +def is_walrus_assignment(node: LN) -> bool: + """Return True iff `node` is of the shape ( test := test )""" + inner = unwrap_singleton_parenthesis(node) + return inner is not None and inner.type == syms.namedexpr_test + + def is_yield(node: LN) -> bool: """Return True if `node` holds a `yield` or `yield from` expression.""" if node.type == syms.yield_expr: @@ -2338,7 +5431,7 @@ def is_vararg(leaf: Leaf, within: Set[NodeType]) -> bool: extended iterable unpacking (PEP 3132) and additional unpacking generalizations (PEP 448). """ - if leaf.type not in STARS or not leaf.parent: + if leaf.type not in VARARGS_SPECIALS or not leaf.parent: return False p = leaf.parent @@ -2353,7 +5446,41 @@ def is_vararg(leaf: Leaf, within: Set[NodeType]) -> bool: return p.type in within -def max_delimiter_priority_in_atom(node: LN) -> int: +def is_multiline_string(leaf: Leaf) -> bool: + """Return True if `leaf` is a multiline string that actually spans many lines.""" + return has_triple_quotes(leaf.value) and "\n" in leaf.value + + +def is_stub_suite(node: Node) -> bool: + """Return True if `node` is a suite with a stub body.""" + if ( + len(node.children) != 4 + or node.children[0].type != token.NEWLINE + or node.children[1].type != token.INDENT + or node.children[3].type != token.DEDENT + ): + return False + + return is_stub_body(node.children[2]) + + +def is_stub_body(node: LN) -> bool: + """Return True if `node` is a simple statement containing an ellipsis.""" + if not isinstance(node, Node) or node.type != syms.simple_stmt: + return False + + if len(node.children) != 2: + return False + + child = node.children[0] + return ( + child.type == syms.atom + and len(child.children) == 3 + and all(leaf == Leaf(token.DOT, ".") for leaf in child.children) + ) + + +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. @@ -2385,7 +5512,7 @@ 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`). + :func:`normalize_invisible_parens` and :func:`visit_import_from`). """ if leaf.type == token.LPAR: leaf.value = "(" @@ -2393,62 +5520,289 @@ def ensure_visible(leaf: Leaf) -> None: leaf.value = ")" -def is_python36(node: Node) -> bool: - """Return True if the current file is using Python 3.6+ features. +def should_explode(line: Line, opening_bracket: Leaf) -> bool: + """Should `line` immediately be split with `delimiter_split()` after RHS?""" + + if not ( + opening_bracket.parent + and opening_bracket.parent.type in {syms.atom, syms.import_from} + and opening_bracket.value in "[{(" + ): + return False + + try: + last_leaf = line.leaves[-1] + exclude = {id(last_leaf)} if last_leaf.type == token.COMMA else set() + max_priority = line.bracket_tracker.max_delimiter_priority(exclude=exclude) + except (IndexError, ValueError): + return False + + return max_priority == COMMA_PRIORITY + + +def get_features_used(node: Node) -> Set[Feature]: + """Return a set of (relatively) new Python features used in this file. Currently looking for: - - f-strings; and - - trailing commas after * or ** in function signatures and calls. + - f-strings; + - underscores in numeric literals; + - trailing commas after * or ** in function signatures and calls; + - positional only arguments in function signatures and lambdas; """ + features: Set[Feature] = set() for n in node.pre_order(): if n.type == token.STRING: value_head = n.value[:2] # type: ignore if value_head in {'f"', 'F"', "f'", "F'", "rf", "fr", "RF", "FR"}: - return True + features.add(Feature.F_STRINGS) + + elif n.type == token.NUMBER: + if "_" in n.value: # type: ignore + features.add(Feature.NUMERIC_UNDERSCORES) + + elif n.type == token.SLASH: + if n.parent and n.parent.type in {syms.typedargslist, syms.arglist}: + features.add(Feature.POS_ONLY_ARGUMENTS) + + elif n.type == token.COLONEQUAL: + features.add(Feature.ASSIGNMENT_EXPRESSIONS) elif ( n.type in {syms.typedargslist, syms.arglist} and n.children and n.children[-1].type == token.COMMA ): + if n.type == syms.typedargslist: + feature = Feature.TRAILING_COMMA_IN_DEF + else: + feature = Feature.TRAILING_COMMA_IN_CALL + for ch in n.children: if ch.type in STARS: - return True + features.add(feature) if ch.type == syms.argument: for argch in ch.children: if argch.type in STARS: - return True + features.add(feature) - return False + return features -PYTHON_EXTENSIONS = {".py"} -BLACKLISTED_DIRECTORIES = { - "build", "buck-out", "dist", "_build", ".git", ".hg", ".mypy_cache", ".tox", ".venv" -} +def detect_target_versions(node: Node) -> Set[TargetVersion]: + """Detect the version to target based on the nodes used.""" + features = get_features_used(node) + return { + version for version in TargetVersion if features <= VERSION_TO_FEATURES[version] + } + + +def generate_trailers_to_omit(line: Line, line_length: int) -> Iterator[Set[LeafID]]: + """Generate sets of closing bracket IDs that should be omitted in a RHS. + + Brackets can be omitted if the entire trailer up to and including + a preceding closing bracket fits in one line. + + Yielded sets are cumulative (contain results of previous yields, too). First + set is empty. + """ + + omit: Set[LeafID] = set() + yield omit + + length = 4 * line.depth + opening_bracket: Optional[Leaf] = None + closing_bracket: Optional[Leaf] = None + inner_brackets: Set[LeafID] = set() + for index, leaf, leaf_length in enumerate_with_length(line, reversed=True): + length += leaf_length + if length > line_length: + break + + has_inline_comment = leaf_length > len(leaf.value) + len(leaf.prefix) + if leaf.type == STANDALONE_COMMENT or has_inline_comment: + break + + if opening_bracket: + if leaf is opening_bracket: + opening_bracket = None + elif leaf.type in CLOSING_BRACKETS: + inner_brackets.add(id(leaf)) + elif leaf.type in CLOSING_BRACKETS: + if index > 0 and line.leaves[index - 1].type in OPENING_BRACKETS: + # Empty brackets would fail a split so treat them as "inner" + # brackets (e.g. only add them to the `omit` set if another + # pair of brackets was good enough. + inner_brackets.add(id(leaf)) + continue + + if closing_bracket: + omit.add(id(closing_bracket)) + omit.update(inner_brackets) + inner_brackets.clear() + yield omit + + if leaf.value: + opening_bracket = leaf.opening_bracket + closing_bracket = leaf + + +def get_future_imports(node: Node) -> Set[str]: + """Return a set of __future__ imports in the file.""" + 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 -def gen_python_files_in_dir(path: Path) -> Iterator[Path]: - """Generate all files under `path` which aren't under BLACKLISTED_DIRECTORIES - and have one of the PYTHON_EXTENSIONS. + elif child.type == syms.import_as_names: + yield from get_imports_from_children(child.children) + + else: + raise AssertionError("Invalid syntax parsing imports") + + for child in node.children: + if child.type != syms.simple_stmt: + break + + first_child = child.children[0] + if isinstance(first_child, Leaf): + # Continue looking if we see a docstring; otherwise stop. + if ( + len(child.children) == 2 + and first_child.type == token.STRING + and child.children[1].type == token.NEWLINE + ): + continue + + break + + elif first_child.type == syms.import_from: + module_name = first_child.children[1] + if not isinstance(module_name, Leaf) or module_name.value != "__future__": + break + + imports |= set(get_imports_from_children(first_child.children[3:])) + else: + break + + return imports + + +@lru_cache() +def get_gitignore(root: Path) -> PathSpec: + """ Return a PathSpec matching gitignore content if present.""" + gitignore = root / ".gitignore" + lines: List[str] = [] + if gitignore.is_file(): + with gitignore.open() as gf: + lines = gf.readlines() + return PathSpec.from_lines("gitwildmatch", lines) + + +def gen_python_files_in_dir( + path: Path, + root: Path, + include: Pattern[str], + exclude: Pattern[str], + report: "Report", + gitignore: PathSpec, +) -> 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(): - if child.is_dir(): - if child.name in BLACKLISTED_DIRECTORIES: + # First ignore files matching .gitignore + if gitignore.match_file(child.as_posix()): + report.path_ignored(child, "matches the .gitignore file content") + continue + + # Then ignore with `exclude` option. + try: + normalized_path = "/" + child.resolve().relative_to(root).as_posix() + except OSError as e: + report.path_ignored(child, f"cannot be read because {e}") + continue + + except ValueError: + if child.is_symlink(): + report.path_ignored( + child, f"is a symbolic link that points outside {root}" + ) continue - yield from gen_python_files_in_dir(child) + 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, "matches the --exclude regular expression") + continue + + if child.is_dir(): + yield from gen_python_files_in_dir( + child, root, include, exclude, report, gitignore + ) + + elif child.is_file(): + include_match = include.search(normalized_path) + if include_match: + yield child + - elif child.suffix in PYTHON_EXTENSIONS: - yield child +@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 + common parent. + + If no directory in the tree contains a marker that would specify it's the + project root, the root of the file system is returned. + """ + if not srcs: + return Path("/").resolve() + + common_base = min(Path(src).resolve() for src in srcs) + if common_base.is_dir(): + # Append a fake file so `parents` below returns `common_base_dir`, too. + common_base /= "fake-file" + for directory in common_base.parents: + if (directory / ".git").exists(): + return directory + + if (directory / ".hg").is_dir(): + return directory + + if (directory / "pyproject.toml").is_file(): + return directory + + return directory @dataclass class Report: """Provides a reformatting counter. Can be rendered with `str(report)`.""" + check: bool = False + diff: bool = False quiet: bool = False + verbose: bool = False change_count: int = 0 same_count: int = 0 failure_count: int = 0 @@ -2456,12 +5810,12 @@ class Report: def done(self, src: Path, changed: Changed) -> None: """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: + reformatted = "would reformat" if self.check or self.diff else "reformatted" + 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: @@ -2474,6 +5828,10 @@ 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. @@ -2484,7 +5842,7 @@ 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 @@ -2498,7 +5856,7 @@ class Report: Use `click.unstyle` to remove colors. """ - if self.check: + if self.check or self.diff: reformatted = "would be reformatted" unchanged = "would be left unchanged" failed = "would fail to reformat" @@ -2523,89 +5881,148 @@ class Report: return ", ".join(report) + "." -def assert_equivalent(src: str, dst: str) -> None: - """Raise AssertionError if `src` and `dst` aren't equivalent.""" +def parse_ast(src: str) -> Union[ast.AST, ast3.AST, ast27.AST]: + filename = "" + if sys.version_info >= (3, 8): + # TODO: support Python 4+ ;) + for minor_version in range(sys.version_info[1], 4, -1): + try: + return ast.parse(src, filename, feature_version=(3, minor_version)) + except SyntaxError: + continue + else: + for feature_version in (7, 6): + try: + return ast3.parse(src, filename, feature_version=feature_version) + except SyntaxError: + continue - import ast - import traceback + return ast27.parse(src) - def _v(node: ast.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): - try: - value = getattr(node, field) - except AttributeError: - continue +def _fixup_ast_constants( + node: Union[ast.AST, ast3.AST, ast27.AST] +) -> Union[ast.AST, ast3.AST, ast27.AST]: + """Map ast nodes deprecated in 3.8 to Constant.""" + if isinstance(node, (ast.Str, ast3.Str, ast27.Str, ast.Bytes, ast3.Bytes)): + return ast.Constant(value=node.s) + + if isinstance(node, (ast.Num, ast3.Num, ast27.Num)): + return ast.Constant(value=node.n) + + if isinstance(node, (ast.NameConstant, ast3.NameConstant)): + return ast.Constant(value=node.value) + + return node + + +def _stringify_ast( + node: Union[ast.AST, ast3.AST, ast27.AST], depth: int = 0 +) -> Iterator[str]: + """Simple visitor generating strings to compare ASTs by content.""" - yield f"{' ' * (depth+1)}{field}=" + node = _fixup_ast_constants(node) - if isinstance(value, list): - for item in value: - if isinstance(item, ast.AST): - yield from _v(item, depth + 2) + yield f"{' ' * depth}{node.__class__.__name__}(" - elif isinstance(value, ast.AST): - yield from _v(value, depth + 2) + for field in sorted(node._fields): # noqa: F402 + # TypeIgnore has only one field 'lineno' which breaks this comparison + type_ignore_classes = (ast3.TypeIgnore, ast27.TypeIgnore) + if sys.version_info >= (3, 8): + type_ignore_classes += (ast.TypeIgnore,) + if isinstance(node, type_ignore_classes): + break + + try: + value = getattr(node, field) + except AttributeError: + continue + + yield f"{' ' * (depth+1)}{field}=" + + if isinstance(value, list): + for item in value: + # Ignore nested tuples within del statements, because we may insert + # parentheses and they change the AST. + if ( + field == "targets" + and isinstance(node, (ast.Delete, ast3.Delete, ast27.Delete)) + and isinstance(item, (ast.Tuple, ast3.Tuple, ast27.Tuple)) + ): + for item in item.elts: + yield from _stringify_ast(item, depth + 2) + + elif isinstance(item, (ast.AST, ast3.AST, ast27.AST)): + yield from _stringify_ast(item, depth + 2) + + elif isinstance(value, (ast.AST, ast3.AST, ast27.AST)): + yield from _stringify_ast(value, depth + 2) + else: + # Constant strings may be indented across newlines, if they are + # docstrings; fold spaces after newlines when comparing + if ( + isinstance(node, ast.Constant) + and field == "value" + and isinstance(value, str) + ): + normalized = re.sub(r"\n[ \t]+", "\n ", value) else: - yield f"{' ' * (depth+2)}{value!r}, # {value.__class__.__name__}" + normalized = value + yield f"{' ' * (depth+2)}{normalized!r}, # {value.__class__.__name__}" + + yield f"{' ' * depth}) # /{node.__class__.__name__}" - yield f"{' ' * depth}) # /{node.__class__.__name__}" +def assert_equivalent(src: str, dst: str) -> None: + """Raise AssertionError if `src` and `dst` aren't equivalent.""" 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}" + "cannot use --safe with this file; failed to parse source file. AST" + f" 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( - f"INTERNAL ERROR: Black produced invalid code: {exc}. " - f"Please report a bug on https://github.com/ambv/black/issues. " - f"This invalid output might be helpful: {log}" + f"INTERNAL ERROR: Black produced invalid code: {exc}. Please report a bug" + " on https://github.com/psf/black/issues. This invalid output might be" + f" helpful: {log}" ) from None - src_ast_str = "\n".join(_v(src_ast)) - dst_ast_str = "\n".join(_v(dst_ast)) + src_ast_str = "\n".join(_stringify_ast(src_ast)) + dst_ast_str = "\n".join(_stringify_ast(dst_ast)) if src_ast_str != dst_ast_str: log = dump_to_file(diff(src_ast_str, dst_ast_str, "src", "dst")) raise AssertionError( - f"INTERNAL ERROR: Black produced code that is not equivalent to " - f"the source. " - f"Please report a bug on https://github.com/ambv/black/issues. " - f"This diff might be helpful: {log}" + "INTERNAL ERROR: Black produced code that is not equivalent to the" + " source. Please report a bug on https://github.com/psf/black/issues. " + f" This diff might be helpful: {log}" ) from None -def assert_stable(src: str, dst: str, line_length: int) -> None: +def assert_stable(src: str, dst: str, mode: Mode) -> None: """Raise AssertionError if `dst` reformats differently the second time.""" - newdst = format_str(dst, line_length=line_length) + newdst = format_str(dst, mode=mode) if dst != newdst: log = dump_to_file( diff(src, dst, "source", "first pass"), diff(dst, newdst, "first pass", "second pass"), ) raise AssertionError( - f"INTERNAL ERROR: Black produced different code on the second pass " - f"of the formatter. " - f"Please report a bug on https://github.com/ambv/black/issues. " - f"This diff might be helpful: {log}" + "INTERNAL ERROR: Black produced different code on the second pass of the" + " formatter. Please report a bug on https://github.com/psf/black/issues." + f" This diff might be helpful: {log}" ) from None +@mypyc_attr(patchable=True) 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: @@ -2616,29 +6033,42 @@ def dump_to_file(*output: str) -> str: return f.name +@contextmanager +def nullcontext() -> Iterator[None]: + """Return an empty context manager. + + To be used like `nullcontext` in Python 3.7. + """ + yield + + def diff(a: str, b: str, a_name: str, b_name: str) -> str: """Return a unified diff string between strings `a` and `b`.""" import difflib - a_lines = [line + "\n" for line in a.split("\n")] - b_lines = [line + "\n" for line in b.split("\n")] + a_lines = [line + "\n" for line in a.splitlines()] + b_lines = [line + "\n" for line in b.splitlines()] return "".join( difflib.unified_diff(a_lines, b_lines, fromfile=a_name, tofile=b_name, n=5) ) -def cancel(tasks: List[asyncio.Task]) -> None: +def cancel(tasks: Iterable["asyncio.Task[Any]"]) -> None: """asyncio signal handler that cancels all `tasks` and reports to stderr.""" err("Aborted!") for task in tasks: task.cancel() -def shutdown(loop: BaseEventLoop) -> None: +def shutdown(loop: asyncio.AbstractEventLoop) -> None: """Cancel all pending tasks on `loop`, wait for them, and close the loop.""" try: + if sys.version_info[:2] >= (3, 7): + all_tasks = asyncio.all_tasks + else: + all_tasks = asyncio.Task.all_tasks # This part is borrowed from asyncio/runners.py in Python 3.7b2. - to_cancel = [task for task in asyncio.Task.all_tasks(loop) if not task.done()] + to_cancel = [task for task in all_tasks(loop) if not task.done()] if not to_cancel: return @@ -2665,26 +6095,203 @@ def sub_twice(regex: Pattern[str], replacement: str, original: str) -> str: return regex.sub(replacement, regex.sub(replacement, original)) -CACHE_DIR = Path(user_cache_dir("black", version=__version__)) +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 + compiled: Pattern[str] = re.compile(regex) + return compiled + + +def enumerate_reversed(sequence: Sequence[T]) -> Iterator[Tuple[Index, T]]: + """Like `reversed(enumerate(sequence))` if that were possible.""" + index = len(sequence) - 1 + for element in reversed(sequence): + yield (index, element) + index -= 1 + + +def enumerate_with_length( + line: Line, reversed: bool = False +) -> Iterator[Tuple[Index, Leaf, int]]: + """Return an enumeration of leaves with their length. + + Stops prematurely on multiline strings and standalone comments. + """ + op = cast( + Callable[[Sequence[Leaf]], Iterator[Tuple[Index, Leaf]]], + enumerate_reversed if reversed else enumerate, + ) + for index, leaf in op(line.leaves): + length = len(leaf.prefix) + len(leaf.value) + if "\n" in leaf.value: + return # Multiline strings, we can't continue. + + for comment in line.comments_after(leaf): + length += len(comment.value) + + yield index, leaf, length -def get_cache_file(line_length: int) -> Path: - return CACHE_DIR / f"cache.{line_length}.pickle" +def is_line_short_enough(line: Line, *, line_length: int, line_str: str = "") -> bool: + """Return True if `line` is no longer than `line_length`. + Uses the provided `line_str` rendering, if any, otherwise computes a new one. + """ + if not line_str: + line_str = line_to_string(line) + return ( + len(line_str) <= line_length + and "\n" not in line_str # multiline strings + and not line.contains_standalone_comments() + ) + + +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? + + Returns True for only a subset of potentially nice looking formattings but + the point is to not return false positives that end up producing lines that + are too long. + """ + bt = line.bracket_tracker + if not bt.delimiters: + # Without delimiters the optional parentheses are useless. + return True + + max_priority = bt.max_delimiter_priority() + if bt.delimiter_count_with_priority(max_priority) > 1: + # With more than one delimiter of a kind the optional parentheses read better. + return False + + if max_priority == DOT_PRIORITY: + # A single stranded method call doesn't require optional parentheses. + return True + + assert len(line.leaves) >= 2, "Stranded delimiter" + + first = line.leaves[0] + second = line.leaves[1] + penultimate = line.leaves[-2] + last = line.leaves[-1] + + # With a single delimiter, omit if the expression starts or ends with + # a bracket. + if first.type in OPENING_BRACKETS and second.type not in CLOSING_BRACKETS: + remainder = False + length = 4 * line.depth + for _index, leaf, leaf_length in enumerate_with_length(line): + if leaf.type in CLOSING_BRACKETS and leaf.opening_bracket is first: + remainder = True + if remainder: + length += leaf_length + if length > line_length: + break + + if leaf.type in OPENING_BRACKETS: + # There are brackets we can further split on. + remainder = False + + else: + # checked the entire string and line length wasn't exceeded + if len(line.leaves) == _index + 1: + return True + + # Note: we are not returning False here because a line might have *both* + # a leading opening bracket and a trailing closing bracket. If the + # opening bracket doesn't match our rule, maybe the closing will. + + if ( + last.type == token.RPAR + or last.type == token.RBRACE + or ( + # don't use indexing for omitting optional parentheses; + # it looks weird + last.type == token.RSQB + and last.parent + and last.parent.type != syms.trailer + ) + ): + if penultimate.type in OPENING_BRACKETS: + # Empty brackets don't help. + return False + + if is_multiline_string(first): + # Additional wrapping of a multiline string in this situation is + # unnecessary. + return True + + length = 4 * line.depth + seen_other_brackets = False + for _index, leaf, leaf_length in enumerate_with_length(line): + length += leaf_length + if leaf is last.opening_bracket: + if seen_other_brackets or length <= line_length: + return True + + elif leaf.type in OPENING_BRACKETS: + # There are brackets we can further split on. + seen_other_brackets = True + + return False + + +def get_cache_file(mode: Mode) -> Path: + return CACHE_DIR / f"cache.{mode.get_cache_key()}.pickle" -def read_cache(line_length: int) -> Cache: + +def read_cache(mode: Mode) -> 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. """ - cache_file = get_cache_file(line_length) + cache_file = get_cache_file(mode) if not cache_file.exists(): return {} with cache_file.open("rb") as fobj: try: cache: Cache = pickle.load(fobj) - except pickle.UnpicklingError: + except (pickle.UnpicklingError, ValueError): return {} return cache @@ -2696,36 +6303,89 @@ 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) -> None: +def write_cache(cache: Cache, sources: Iterable[Path], mode: Mode) -> None: """Update the cache file.""" - cache_file = get_cache_file(line_length) + cache_file = get_cache_file(mode) try: - if not CACHE_DIR.exists(): - CACHE_DIR.mkdir(parents=True) + CACHE_DIR.mkdir(parents=True, exist_ok=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) + with tempfile.NamedTemporaryFile(dir=str(cache_file.parent), delete=False) as f: + pickle.dump(new_cache, f, protocol=4) + os.replace(f.name, cache_file) except OSError: pass -if __name__ == "__main__": +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 + + +def patched_main() -> None: + freeze_support() + patch_click() main() + + +def fix_docstring(docstring: str, prefix: str) -> str: + # https://www.python.org/dev/peps/pep-0257/#handling-docstring-indentation + if not docstring: + return "" + # Convert tabs to spaces (following the normal Python rules) + # and split into a list of lines: + lines = docstring.expandtabs().splitlines() + # Determine minimum indentation (first line doesn't count): + indent = sys.maxsize + for line in lines[1:]: + stripped = line.lstrip() + if stripped: + indent = min(indent, len(line) - len(stripped)) + # Remove indentation (first line is special): + trimmed = [lines[0].strip()] + if indent < sys.maxsize: + last_line_idx = len(lines) - 2 + for i, line in enumerate(lines[1:]): + stripped_line = line[indent:].rstrip() + if stripped_line or i == last_line_idx: + trimmed.append(prefix + stripped_line) + else: + trimmed.append("") + # Return a single string: + return "\n".join(trimmed) + + +if __name__ == "__main__": + patched_main()