src/black/files.py

   1 from functools import lru_cache
   2 import io
   3 import os
   4 from pathlib import Path
   5 import sys
   6 from typing import (
   7     Any,
   8     Dict,
   9     Iterable,
  10     Iterator,
  11     List,
  12     Optional,
  13     Pattern,
  14     Sequence,
  15     Tuple,
  16     Union,
  17     TYPE_CHECKING,
  18 )
  19
  20 from mypy_extensions import mypyc_attr
  21 from pathspec import PathSpec
  22 from pathspec.patterns.gitwildmatch import GitWildMatchPatternError
  23 import tomli
  24
  25 from black.output import err
  26 from black.report import Report
  27 from black.handle_ipynb_magics import jupyter_dependencies_are_installed
  28
  29 if TYPE_CHECKING:
  30     import colorama  # noqa: F401
  31
  32
  33 @lru_cache()
  34 def find_project_root(srcs: Sequence[str]) -> Path:
  35     """Return a directory containing .git, .hg, or pyproject.toml.
  36
  37     That directory will be a common parent of all files and directories
  38     passed in `srcs`.
  39
  40     If no directory in the tree contains a marker that would specify it's the
  41     project root, the root of the file system is returned.
  42     """
  43     if not srcs:
  44         srcs = [str(Path.cwd().resolve())]
  45
  46     path_srcs = [Path(Path.cwd(), src).resolve() for src in srcs]
  47
  48     # A list of lists of parents for each 'src'. 'src' is included as a
  49     # "parent" of itself if it is a directory
  50     src_parents = [
  51         list(path.parents) + ([path] if path.is_dir() else []) for path in path_srcs
  52     ]
  53
  54     common_base = max(
  55         set.intersection(*(set(parents) for parents in src_parents)),
  56         key=lambda path: path.parts,
  57     )
  58
  59     for directory in (common_base, *common_base.parents):
  60         if (directory / ".git").exists():
  61             return directory
  62
  63         if (directory / ".hg").is_dir():
  64             return directory
  65
  66         if (directory / "pyproject.toml").is_file():
  67             return directory
  68
  69     return directory
  70
  71
  72 def find_pyproject_toml(path_search_start: Tuple[str, ...]) -> Optional[str]:
  73     """Find the absolute filepath to a pyproject.toml if it exists"""
  74     path_project_root = find_project_root(path_search_start)
  75     path_pyproject_toml = path_project_root / "pyproject.toml"
  76     if path_pyproject_toml.is_file():
  77         return str(path_pyproject_toml)
  78
  79     try:
  80         path_user_pyproject_toml = find_user_pyproject_toml()
  81         return (
  82             str(path_user_pyproject_toml)
  83             if path_user_pyproject_toml.is_file()
  84             else None
  85         )
  86     except PermissionError as e:
  87         # We do not have access to the user-level config directory, so ignore it.
  88         err(f"Ignoring user configuration directory due to {e!r}")
  89         return None
  90
  91
  92 @mypyc_attr(patchable=True)
  93 def parse_pyproject_toml(path_config: str) -> Dict[str, Any]:
  94     """Parse a pyproject toml file, pulling out relevant parts for Black
  95
  96     If parsing fails, will raise a tomli.TOMLDecodeError
  97     """
  98     with open(path_config, "rb") as f:
  99         pyproject_toml = tomli.load(f)
 100     config = pyproject_toml.get("tool", {}).get("black", {})
 101     return {k.replace("--", "").replace("-", "_"): v for k, v in config.items()}
 102
 103
 104 @lru_cache()
 105 def find_user_pyproject_toml() -> Path:
 106     r"""Return the path to the top-level user configuration for black.
 107
 108     This looks for ~\.black on Windows and ~/.config/black on Linux and other
 109     Unix systems.
 110     """
 111     if sys.platform == "win32":
 112         # Windows
 113         user_config_path = Path.home() / ".black"
 114     else:
 115         config_root = os.environ.get("XDG_CONFIG_HOME", "~/.config")
 116         user_config_path = Path(config_root).expanduser() / "black"
 117     return user_config_path.resolve()
 118
 119
 120 @lru_cache()
 121 def get_gitignore(root: Path) -> PathSpec:
 122     """Return a PathSpec matching gitignore content if present."""
 123     gitignore = root / ".gitignore"
 124     lines: List[str] = []
 125     if gitignore.is_file():
 126         with gitignore.open(encoding="utf-8") as gf:
 127             lines = gf.readlines()
 128     try:
 129         return PathSpec.from_lines("gitwildmatch", lines)
 130     except GitWildMatchPatternError as e:
 131         err(f"Could not parse {gitignore}: {e}")
 132         raise
 133
 134
 135 def normalize_path_maybe_ignore(
 136     path: Path, root: Path, report: Report
 137 ) -> Optional[str]:
 138     """Normalize `path`. May return `None` if `path` was ignored.
 139
 140     `report` is where "path ignored" output goes.
 141     """
 142     try:
 143         abspath = path if path.is_absolute() else Path.cwd() / path
 144         normalized_path = abspath.resolve().relative_to(root).as_posix()
 145     except OSError as e:
 146         report.path_ignored(path, f"cannot be read because {e}")
 147         return None
 148
 149     except ValueError:
 150         if path.is_symlink():
 151             report.path_ignored(path, f"is a symbolic link that points outside {root}")
 152             return None
 153
 154         raise
 155
 156     return normalized_path
 157
 158
 159 def path_is_excluded(
 160     normalized_path: str,
 161     pattern: Optional[Pattern[str]],
 162 ) -> bool:
 163     match = pattern.search(normalized_path) if pattern else None
 164     return bool(match and match.group(0))
 165
 166
 167 def gen_python_files(
 168     paths: Iterable[Path],
 169     root: Path,
 170     include: Pattern[str],
 171     exclude: Pattern[str],
 172     extend_exclude: Optional[Pattern[str]],
 173     force_exclude: Optional[Pattern[str]],
 174     report: Report,
 175     gitignore: Optional[PathSpec],
 176     *,
 177     verbose: bool,
 178     quiet: bool,
 179 ) -> Iterator[Path]:
 180     """Generate all files under `path` whose paths are not excluded by the
 181     `exclude_regex`, `extend_exclude`, or `force_exclude` regexes,
 182     but are included by the `include` regex.
 183
 184     Symbolic links pointing outside of the `root` directory are ignored.
 185
 186     `report` is where output about exclusions goes.
 187     """
 188     assert root.is_absolute(), f"INTERNAL ERROR: `root` must be absolute but is {root}"
 189     for child in paths:
 190         normalized_path = normalize_path_maybe_ignore(child, root, report)
 191         if normalized_path is None:
 192             continue
 193
 194         # First ignore files matching .gitignore, if passed
 195         if gitignore is not None and gitignore.match_file(normalized_path):
 196             report.path_ignored(child, "matches the .gitignore file content")
 197             continue
 198
 199         # Then ignore with `--exclude` `--extend-exclude` and `--force-exclude` options.
 200         normalized_path = "/" + normalized_path
 201         if child.is_dir():
 202             normalized_path += "/"
 203
 204         if path_is_excluded(normalized_path, exclude):
 205             report.path_ignored(child, "matches the --exclude regular expression")
 206             continue
 207
 208         if path_is_excluded(normalized_path, extend_exclude):
 209             report.path_ignored(
 210                 child, "matches the --extend-exclude regular expression"
 211             )
 212             continue
 213
 214         if path_is_excluded(normalized_path, force_exclude):
 215             report.path_ignored(child, "matches the --force-exclude regular expression")
 216             continue
 217
 218         if child.is_dir():
 219             # If gitignore is None, gitignore usage is disabled, while a Falsey
 220             # gitignore is when the directory doesn't have a .gitignore file.
 221             yield from gen_python_files(
 222                 child.iterdir(),
 223                 root,
 224                 include,
 225                 exclude,
 226                 extend_exclude,
 227                 force_exclude,
 228                 report,
 229                 gitignore + get_gitignore(child) if gitignore is not None else None,
 230                 verbose=verbose,
 231                 quiet=quiet,
 232             )
 233
 234         elif child.is_file():
 235             if child.suffix == ".ipynb" and not jupyter_dependencies_are_installed(
 236                 verbose=verbose, quiet=quiet
 237             ):
 238                 continue
 239             include_match = include.search(normalized_path) if include else True
 240             if include_match:
 241                 yield child
 242
 243
 244 def wrap_stream_for_windows(
 245     f: io.TextIOWrapper,
 246 ) -> Union[io.TextIOWrapper, "colorama.AnsiToWin32"]:
 247     """
 248     Wrap stream with colorama's wrap_stream so colors are shown on Windows.
 249
 250     If `colorama` is unavailable, the original stream is returned unmodified.
 251     Otherwise, the `wrap_stream()` function determines whether the stream needs
 252     to be wrapped for a Windows environment and will accordingly either return
 253     an `AnsiToWin32` wrapper or the original stream.
 254     """
 255     try:
 256         from colorama.initialise import wrap_stream
 257     except ImportError:
 258         return f
 259     else:
 260         # Set `strip=False` to avoid needing to modify test_express_diff_with_color.
 261         return wrap_stream(f, convert=None, strip=False, autoreset=False, wrap=True)