]> git.madduck.net Git - etc/vim.git/blobdiff - src/black/__init__.py

madduck's git repository

Every one of the projects in this repository is available at the canonical URL git://git.madduck.net/madduck/pub/<projectpath> — see each project's metadata for the exact URL.

All patches and comments are welcome. Please squash your changes to logical commits before using git-format-patch and git-send-email to patches@git.madduck.net. If you'd read over the Git project's submission guidelines and adhered to them, I'd be especially grateful.

SSH access, as well as push access can be individually arranged.

If you use my repositories frequently, consider adding the following snippet to ~/.gitconfig and using the third clone URL listed for each project:

[url "git://git.madduck.net/madduck/"]
  insteadOf = madduck:

Bump docker/setup-qemu-action from 1 to 2 (#3056)
[etc/vim.git] / src / black / __init__.py
index fdbaf040d642b0a41d15d0f98112090971425643..75321c3f35c96c2ebfc2e62d755ac0500ad16a73 100644 (file)
@@ -1,7 +1,6 @@
 import asyncio
 from json.decoder import JSONDecodeError
 import json
-from concurrent.futures import Executor, ThreadPoolExecutor, ProcessPoolExecutor
 from contextlib import contextmanager
 from datetime import datetime
 from enum import Enum
@@ -10,12 +9,14 @@ from multiprocessing import Manager, freeze_support
 import os
 from pathlib import Path
 from pathspec.patterns.gitwildmatch import GitWildMatchPatternError
-import regex as re
+import platform
+import re
 import signal
 import sys
 import tokenize
 import traceback
 from typing import (
+    TYPE_CHECKING,
     Any,
     Dict,
     Generator,
@@ -24,28 +25,37 @@ from typing import (
     MutableMapping,
     Optional,
     Pattern,
+    Sequence,
     Set,
     Sized,
     Tuple,
     Union,
 )
 
-from dataclasses import replace
 import click
+from click.core import ParameterSource
+from dataclasses import replace
+from mypy_extensions import mypyc_attr
 
 from black.const import DEFAULT_LINE_LENGTH, DEFAULT_INCLUDES, DEFAULT_EXCLUDES
 from black.const import STDIN_PLACEHOLDER
 from black.nodes import STARS, syms, is_simple_decorator_expression
+from black.nodes import is_string_token
 from black.lines import Line, EmptyLineTracker
 from black.linegen import transform_line, LineGenerator, LN
 from black.comments import normalize_fmt_off
-from black.mode import Mode, TargetVersion
+from black.mode import FUTURE_FLAG_TO_FEATURE, Mode, TargetVersion
 from black.mode import Feature, supports_feature, VERSION_TO_FEATURES
 from black.cache import read_cache, write_cache, get_cache_info, filter_cached, Cache
 from black.concurrency import cancel, shutdown, maybe_install_uvloop
 from black.output import dump_to_file, ipynb_diff, diff, color_diff, out, err
 from black.report import Report, Changed, NothingChanged
-from black.files import find_project_root, find_pyproject_toml, parse_pyproject_toml
+from black.files import (
+    find_project_root,
+    find_pyproject_toml,
+    parse_pyproject_toml,
+    find_user_pyproject_toml,
+)
 from black.files import gen_python_files, get_gitignore, normalize_path_maybe_ignore
 from black.files import wrap_stream_for_windows
 from black.parsing import InvalidInput  # noqa F401
@@ -56,6 +66,7 @@ from black.handle_ipynb_magics import (
     remove_trailing_semicolon,
     put_trailing_semicolon_back,
     TRANSFORMED_MAGICS,
+    PYTHON_CELL_MAGICS,
     jupyter_dependencies_are_installed,
 )
 
@@ -66,6 +77,11 @@ from blib2to3.pgen2 import token
 
 from _black_version import version as __version__
 
+if TYPE_CHECKING:
+    from concurrent.futures import Executor
+
+COMPILED = Path(__file__).suffix in (".pyd", ".so")
+
 # types
 FileContent = str
 Encoding = str
@@ -170,14 +186,19 @@ def validate_regex(
     ctx: click.Context,
     param: click.Parameter,
     value: Optional[str],
-) -> Optional[Pattern]:
+) -> Optional[Pattern[str]]:
     try:
         return re_compile_maybe_verbose(value) if value is not None else None
-    except re.error:
-        raise click.BadParameter("Not a valid regular expression") from None
+    except re.error as e:
+        raise click.BadParameter(f"Not a valid regular expression: {e}") from None
 
 
-@click.command(context_settings=dict(help_option_names=["-h", "--help"]))
+@click.command(
+    context_settings={"help_option_names": ["-h", "--help"]},
+    # While Click does set this field automatically using the docstring, mypyc
+    # (annoyingly) strips 'em so we need to set it here too.
+    help="The uncompromising code formatter.",
+)
 @click.option("-c", "--code", type=str, help="Format the code passed in as a string.")
 @click.option(
     "-l",
@@ -214,6 +235,16 @@ def validate_regex(
         "(useful when piping source on standard input)."
     ),
 )
+@click.option(
+    "--python-cell-magics",
+    multiple=True,
+    help=(
+        "When processing Jupyter Notebooks, add the given magic to the list"
+        f" of known python-magics ({', '.join(PYTHON_CELL_MAGICS)})."
+        " Useful for formatting cells with custom python magics."
+    ),
+    default=[],
+)
 @click.option(
     "-S",
     "--skip-string-normalization",
@@ -230,9 +261,14 @@ def validate_regex(
     "--experimental-string-processing",
     is_flag=True,
     hidden=True,
+    help="(DEPRECATED and now included in --preview) Normalize string literals.",
+)
+@click.option(
+    "--preview",
+    is_flag=True,
     help=(
-        "Experimental option that performs more normalization on string literals."
-        " Currently disabled because it leads to some crashes."
+        "Enable potentially disruptive style changes that may be added to Black's main"
+        " functionality in the next major release."
     ),
 )
 @click.option(
@@ -264,7 +300,8 @@ def validate_regex(
     type=str,
     help=(
         "Require a specific version of Black to be running (useful for unifying results"
-        " across many environments e.g. with a pyproject.toml file)."
+        " across many environments e.g. with a pyproject.toml file). It can be"
+        " either a major version number or an exact version."
     ),
 )
 @click.option(
@@ -346,7 +383,13 @@ def validate_regex(
         " due to exclusion patterns."
     ),
 )
-@click.version_option(version=__version__)
+@click.version_option(
+    version=__version__,
+    message=(
+        f"%(prog)s, %(version)s (compiled: {'yes' if COMPILED else 'no'})\n"
+        f"Python ({platform.python_implementation()}) {platform.python_version()}"
+    ),
+)
 @click.argument(
     "src",
     nargs=-1,
@@ -371,7 +414,7 @@ def validate_regex(
     help="Read configuration from FILE path.",
 )
 @click.pass_context
-def main(
+def main(  # noqa: C901
     ctx: click.Context,
     code: Optional[str],
     line_length: int,
@@ -382,27 +425,83 @@ def main(
     fast: bool,
     pyi: bool,
     ipynb: bool,
+    python_cell_magics: Sequence[str],
     skip_string_normalization: bool,
     skip_magic_trailing_comma: bool,
     experimental_string_processing: bool,
+    preview: bool,
     quiet: bool,
     verbose: bool,
-    required_version: str,
-    include: Pattern,
-    exclude: Optional[Pattern],
-    extend_exclude: Optional[Pattern],
-    force_exclude: Optional[Pattern],
+    required_version: Optional[str],
+    include: Pattern[str],
+    exclude: Optional[Pattern[str]],
+    extend_exclude: Optional[Pattern[str]],
+    force_exclude: Optional[Pattern[str]],
     stdin_filename: Optional[str],
     workers: int,
     src: Tuple[str, ...],
     config: Optional[str],
 ) -> None:
     """The uncompromising code formatter."""
-    if config and verbose:
-        out(f"Using configuration from {config}.", bold=False, fg="blue")
+    ctx.ensure_object(dict)
+
+    if src and code is not None:
+        out(
+            main.get_usage(ctx)
+            + "\n\n'SRC' and 'code' cannot be passed simultaneously."
+        )
+        ctx.exit(1)
+    if not src and code is None:
+        out(main.get_usage(ctx) + "\n\nOne of 'SRC' or 'code' is required.")
+        ctx.exit(1)
+
+    root, method = find_project_root(src) if code is None else (None, None)
+    ctx.obj["root"] = root
+
+    if verbose:
+        if root:
+            out(
+                f"Identified `{root}` as project root containing a {method}.",
+                fg="blue",
+            )
+
+            normalized = [
+                (normalize_path_maybe_ignore(Path(source), root), source)
+                for source in src
+            ]
+            srcs_string = ", ".join(
+                [
+                    f'"{_norm}"'
+                    if _norm
+                    else f'\033[31m"{source} (skipping - invalid)"\033[34m'
+                    for _norm, source in normalized
+                ]
+            )
+            out(f"Sources to be formatted: {srcs_string}", fg="blue")
+
+        if config:
+            config_source = ctx.get_parameter_source("config")
+            user_level_config = str(find_user_pyproject_toml())
+            if config == user_level_config:
+                out(
+                    "Using configuration from user-level config at "
+                    f"'{user_level_config}'.",
+                    fg="blue",
+                )
+            elif config_source in (
+                ParameterSource.DEFAULT,
+                ParameterSource.DEFAULT_MAP,
+            ):
+                out("Using configuration from project root.", fg="blue")
+            else:
+                out(f"Using configuration in '{config}'.", fg="blue")
 
     error_msg = "Oh no! 💥 💔 💥"
-    if required_version and required_version != __version__:
+    if (
+        required_version
+        and required_version != __version__
+        and required_version != __version__.split(".")[0]
+    ):
         err(
             f"{error_msg} The required version `{required_version}` does not match"
             f" the running version `{__version__}`!"
@@ -426,6 +525,8 @@ def main(
         string_normalization=not skip_string_normalization,
         magic_trailing_comma=not skip_magic_trailing_comma,
         experimental_string_processing=experimental_string_processing,
+        preview=preview,
+        python_cell_magics=set(python_cell_magics),
     )
 
     if code is not None:
@@ -483,6 +584,8 @@ def main(
             )
 
     if verbose or not quiet:
+        if code is None and (verbose or report.change_count or report.failure_count):
+            out()
         out(error_msg if report.return_code else "All done! ✨ 🍰 ✨")
         if code is None:
             click.echo(str(report), err=True)
@@ -503,14 +606,11 @@ def get_sources(
     stdin_filename: Optional[str],
 ) -> Set[Path]:
     """Compute the set of files to be formatted."""
-
-    root = find_project_root(src)
     sources: Set[Path] = set()
-    path_empty(src, "No Path provided. Nothing to do 😴", quiet, verbose, ctx)
 
     if exclude is None:
         exclude = re_compile_maybe_verbose(DEFAULT_EXCLUDES)
-        gitignore = get_gitignore(root)
+        gitignore = get_gitignore(ctx.obj["root"])
     else:
         gitignore = None
 
@@ -523,7 +623,7 @@ def get_sources(
             is_stdin = False
 
         if is_stdin or p.is_file():
-            normalized_path = normalize_path_maybe_ignore(p, root, report)
+            normalized_path = normalize_path_maybe_ignore(p, ctx.obj["root"], report)
             if normalized_path is None:
                 continue
 
@@ -550,7 +650,7 @@ def get_sources(
             sources.update(
                 gen_python_files(
                     p.iterdir(),
-                    root,
+                    ctx.obj["root"],
                     include,
                     exclude,
                     extend_exclude,
@@ -604,6 +704,9 @@ def reformat_code(
         report.failed(path, str(exc))
 
 
+# diff-shades depends on being to monkeypatch this function to operate. I know it's
+# not ideal, but this shouldn't cause any issues ... hopefully. ~ichard26
+@mypyc_attr(patchable=True)
 def reformat_one(
     src: Path, fast: bool, write_back: WriteBack, mode: Mode, report: "Report"
 ) -> None:
@@ -655,6 +758,9 @@ def reformat_one(
         report.failed(src, str(exc))
 
 
+# diff-shades depends on being to monkeypatch this function to operate. I know it's
+# not ideal, but this shouldn't cause any issues ... hopefully. ~ichard26
+@mypyc_attr(patchable=True)
 def reformat_many(
     sources: Set[Path],
     fast: bool,
@@ -664,15 +770,18 @@ def reformat_many(
     workers: Optional[int],
 ) -> None:
     """Reformat multiple files using a ProcessPoolExecutor."""
+    from concurrent.futures import Executor, ThreadPoolExecutor, ProcessPoolExecutor
+
     executor: Executor
     loop = asyncio.get_event_loop()
     worker_count = workers if workers is not None else DEFAULT_WORKERS
     if sys.platform == "win32":
         # Work around https://bugs.python.org/issue26903
+        assert worker_count is not None
         worker_count = min(worker_count, 60)
     try:
         executor = ProcessPoolExecutor(max_workers=worker_count)
-    except (ImportError, OSError):
+    except (ImportError, NotImplementedError, OSError):
         # we arrive here if the underlying system does not support multi-processing
         # like in AWS Lambda or Termux, in which case we gracefully fallback to
         # a ThreadPoolExecutor with just a single worker (more workers would not do us
@@ -704,7 +813,7 @@ async def schedule_formatting(
     mode: Mode,
     report: "Report",
     loop: asyncio.AbstractEventLoop,
-    executor: Executor,
+    executor: "Executor",
 ) -> None:
     """Run formatting of `sources` in parallel using the provided `executor`.
 
@@ -763,7 +872,10 @@ async def schedule_formatting(
                     sources_to_cache.append(src)
                 report.done(src, changed)
     if cancelled:
-        await asyncio.gather(*cancelled, loop=loop, return_exceptions=True)
+        if sys.version_info >= (3, 7):
+            await asyncio.gather(*cancelled, return_exceptions=True)
+        else:
+            await asyncio.gather(*cancelled, loop=loop, return_exceptions=True)
     if sources_to_cache:
         write_cache(cache, sources_to_cache, mode)
 
@@ -888,17 +1000,7 @@ def check_stability_and_equivalence(
     content differently.
     """
     assert_equivalent(src_contents, dst_contents)
-
-    # Forced second pass to work around optional trailing commas (becoming
-    # forced trailing commas on pass 2) interacting differently with optional
-    # parentheses.  Admittedly ugly.
-    dst_contents_pass2 = format_str(dst_contents, mode=mode)
-    if dst_contents != dst_contents_pass2:
-        dst_contents = dst_contents_pass2
-        assert_equivalent(src_contents, dst_contents, pass_num=2)
-        assert_stable(src_contents, dst_contents, mode=mode)
-    # Note: no need to explicitly call `assert_stable` if `dst_contents` was
-    # the same as `dst_contents_pass2`.
+    assert_stable(src_contents, dst_contents, mode=mode)
 
 
 def format_file_contents(src_contents: str, *, fast: bool, mode: Mode) -> FileContent:
@@ -924,8 +1026,10 @@ def format_file_contents(src_contents: str, *, fast: bool, mode: Mode) -> FileCo
     return dst_contents
 
 
-def validate_cell(src: str) -> None:
-    """Check that cell does not already contain TransformerManager transformations.
+def validate_cell(src: str, mode: Mode) -> None:
+    """Check that cell does not already contain TransformerManager transformations,
+    or non-Python cell magics, which might cause tokenizer_rt to break because of
+    indentations.
 
     If a cell contains ``!ls``, then it'll be transformed to
     ``get_ipython().system('ls')``. However, if the cell originally contained
@@ -941,6 +1045,11 @@ def validate_cell(src: str) -> None:
     """
     if any(transformed_magic in src for transformed_magic in TRANSFORMED_MAGICS):
         raise NothingChanged
+    if (
+        src[:2] == "%%"
+        and src.split()[0][2:] not in PYTHON_CELL_MAGICS | mode.python_cell_magics
+    ):
+        raise NothingChanged
 
 
 def format_cell(src: str, *, fast: bool, mode: Mode) -> str:
@@ -959,7 +1068,7 @@ def format_cell(src: str, *, fast: bool, mode: Mode) -> str:
     could potentially be automagics or multi-line magics, which
     are currently not supported.
     """
-    validate_cell(src)
+    validate_cell(src, mode)
     src_without_trailing_semicolon, has_trailing_semicolon = remove_trailing_semicolon(
         src
     )
@@ -1021,7 +1130,7 @@ def format_ipynb_string(src_contents: str, *, fast: bool, mode: Mode) -> FileCon
         raise NothingChanged
 
 
-def format_str(src_contents: str, *, mode: Mode) -> FileContent:
+def format_str(src_contents: str, *, mode: Mode) -> str:
     """Reformat a string and return new contents.
 
     `mode` determines formatting options, such as how many characters per line are
@@ -1051,19 +1160,26 @@ def format_str(src_contents: str, *, mode: Mode) -> FileContent:
         hey
 
     """
+    dst_contents = _format_str_once(src_contents, mode=mode)
+    # Forced second pass to work around optional trailing commas (becoming
+    # forced trailing commas on pass 2) interacting differently with optional
+    # parentheses.  Admittedly ugly.
+    if src_contents != dst_contents:
+        return _format_str_once(dst_contents, mode=mode)
+    return dst_contents
+
+
+def _format_str_once(src_contents: str, *, mode: Mode) -> str:
     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(
-        mode=mode,
-        remove_u_prefix="unicode_literals" in future_imports
-        or supports_feature(versions, Feature.UNICODE_LITERALS),
-    )
+        versions = detect_target_versions(src_node, future_imports=future_imports)
+
+    normalize_fmt_off(src_node, preview=mode.preview)
+    lines = LineGenerator(mode=mode)
     elt = EmptyLineTracker(is_pyi=mode.is_pyi)
     empty_line = Line(mode=mode)
     after = 0
@@ -1100,7 +1216,9 @@ def decode_bytes(src: bytes) -> Tuple[FileContent, Encoding, NewLine]:
         return tiow.read(), encoding, newline
 
 
-def get_features_used(node: Node) -> Set[Feature]:
+def get_features_used(  # noqa: C901
+    node: Node, *, future_imports: Optional[Set[str]] = None
+) -> Set[Feature]:
     """Return a set of (relatively) new Python features used in this file.
 
     Currently looking for:
@@ -1110,16 +1228,26 @@ def get_features_used(node: Node) -> Set[Feature]:
     - positional only arguments in function signatures and lambdas;
     - assignment expression;
     - relaxed decorator syntax;
+    - usage of __future__ flags (annotations);
+    - print / exec statements;
     """
     features: Set[Feature] = set()
+    if future_imports:
+        features |= {
+            FUTURE_FLAG_TO_FEATURE[future_import]
+            for future_import in future_imports
+            if future_import in FUTURE_FLAG_TO_FEATURE
+        }
+
     for n in node.pre_order():
-        if n.type == token.STRING:
-            value_head = n.value[:2]  # type: ignore
+        if is_string_token(n):
+            value_head = n.value[:2]
             if value_head in {'f"', 'F"', "f'", "F'", "rf", "fr", "RF", "FR"}:
                 features.add(Feature.F_STRINGS)
 
         elif n.type == token.NUMBER:
-            if "_" in n.value:  # type: ignore
+            assert isinstance(n, Leaf)
+            if "_" in n.value:
                 features.add(Feature.NUMERIC_UNDERSCORES)
 
         elif n.type == token.SLASH:
@@ -1158,12 +1286,36 @@ def get_features_used(node: Node) -> Set[Feature]:
                         if argch.type in STARS:
                             features.add(feature)
 
+        elif (
+            n.type in {syms.return_stmt, syms.yield_expr}
+            and len(n.children) >= 2
+            and n.children[1].type == syms.testlist_star_expr
+            and any(child.type == syms.star_expr for child in n.children[1].children)
+        ):
+            features.add(Feature.UNPACKING_ON_FLOW)
+
+        elif (
+            n.type == syms.annassign
+            and len(n.children) >= 4
+            and n.children[3].type == syms.testlist_star_expr
+        ):
+            features.add(Feature.ANN_ASSIGN_EXTENDED_RHS)
+
+        elif (
+            n.type == syms.except_clause
+            and len(n.children) >= 2
+            and n.children[1].type == token.STAR
+        ):
+            features.add(Feature.EXCEPT_STAR)
+
     return features
 
 
-def detect_target_versions(node: Node) -> Set[TargetVersion]:
+def detect_target_versions(
+    node: Node, *, future_imports: Optional[Set[str]] = None
+) -> Set[TargetVersion]:
     """Detect the version to target based on the nodes used."""
-    features = get_features_used(node)
+    features = get_features_used(node, future_imports=future_imports)
     return {
         version for version in TargetVersion if features <= VERSION_TO_FEATURES[version]
     }
@@ -1219,13 +1371,16 @@ def get_future_imports(node: Node) -> Set[str]:
     return imports
 
 
-def assert_equivalent(src: str, dst: str, *, pass_num: int = 1) -> None:
+def assert_equivalent(src: str, dst: str) -> None:
     """Raise AssertionError if `src` and `dst` aren't equivalent."""
     try:
         src_ast = parse_ast(src)
     except Exception as exc:
         raise AssertionError(
-            "cannot use --safe with this file; failed to parse source file."
+            "cannot use --safe with this file; failed to parse source file AST: "
+            f"{exc}\n"
+            "This could be caused by running Black with an older Python version "
+            "that does not support new syntax used in your source file."
         ) from exc
 
     try:
@@ -1233,7 +1388,7 @@ def assert_equivalent(src: str, dst: str, *, pass_num: int = 1) -> None:
     except Exception as exc:
         log = dump_to_file("".join(traceback.format_tb(exc.__traceback__)), dst)
         raise AssertionError(
-            f"INTERNAL ERROR: Black produced invalid code on pass {pass_num}: {exc}. "
+            f"INTERNAL ERROR: Black produced invalid code: {exc}. "
             "Please report a bug on https://github.com/psf/black/issues.  "
             f"This invalid output might be helpful: {log}"
         ) from None
@@ -1244,14 +1399,17 @@ def assert_equivalent(src: str, dst: str, *, pass_num: int = 1) -> None:
         log = dump_to_file(diff(src_ast_str, dst_ast_str, "src", "dst"))
         raise AssertionError(
             "INTERNAL ERROR: Black produced code that is not equivalent to the"
-            f" source on pass {pass_num}.  Please report a bug on "
+            " source.  Please report a bug on "
             f"https://github.com/psf/black/issues.  This diff might be helpful: {log}"
         ) from None
 
 
 def assert_stable(src: str, dst: str, mode: Mode) -> None:
     """Raise AssertionError if `dst` reformats differently the second time."""
-    newdst = format_str(dst, mode=mode)
+    # We shouldn't call format_str() here, because that formats the string
+    # twice and may hide a bug where we bounce back and forth between two
+    # versions.
+    newdst = _format_str_once(dst, mode=mode)
     if dst != newdst:
         log = dump_to_file(
             str(mode),
@@ -1285,13 +1443,23 @@ def patch_click() -> None:
     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.
     """
+    modules: List[Any] = []
     try:
         from click import core
+    except ImportError:
+        pass
+    else:
+        modules.append(core)
+    try:
+        # Removed in Click 8.1.0 and newer; we keep this around for users who have
+        # older versions installed.
         from click import _unicodefun  # type: ignore
-    except ModuleNotFoundError:
-        return
+    except ImportError:
+        pass
+    else:
+        modules.append(_unicodefun)
 
-    for module in (core, _unicodefun):
+    for module in modules:
         if hasattr(module, "_verify_python3_env"):
             module._verify_python3_env = lambda: None  # type: ignore
         if hasattr(module, "_verify_python_env"):