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]) -> Tuple[Path, str]:
  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     Returns a two-tuple with the first element as the project root path and
  44     the second element as a string describing the method by which the
  45     project root was discovered.
  46     """
  47     if not srcs:
  48         srcs = [str(Path.cwd().resolve())]
  49
  50     path_srcs = [Path(Path.cwd(), src).resolve() for src in srcs]
  51
  52     # A list of lists of parents for each 'src'. 'src' is included as a
  53     # "parent" of itself if it is a directory
  54     src_parents = [
  55         list(path.parents) + ([path] if path.is_dir() else []) for path in path_srcs
  56     ]
  57
  58     common_base = max(
  59         set.intersection(*(set(parents) for parents in src_parents)),
  60         key=lambda path: path.parts,
  61     )
  62
  63     for directory in (common_base, *common_base.parents):
  64         if (directory / ".git").exists():
  65             return directory, ".git directory"
  66
  67         if (directory / ".hg").is_dir():
  68             return directory, ".hg directory"
  69
  70         if (directory / "pyproject.toml").is_file():
  71             return directory, "pyproject.toml"
  72
  73     return directory, "file system root"
  74
  75
  76 def find_pyproject_toml(path_search_start: Tuple[str, ...]) -> Optional[str]:
  77     """Find the absolute filepath to a pyproject.toml if it exists"""
  78     path_project_root, _ = find_project_root(path_search_start)
  79     path_pyproject_toml = path_project_root / "pyproject.toml"
  80     if path_pyproject_toml.is_file():
  81         return str(path_pyproject_toml)
  82
  83     try:
  84         path_user_pyproject_toml = find_user_pyproject_toml()
  85         return (
  86             str(path_user_pyproject_toml)
  87             if path_user_pyproject_toml.is_file()
  88             else None
  89         )
  90     except PermissionError as e:
  91         # We do not have access to the user-level config directory, so ignore it.
  92         err(f"Ignoring user configuration directory due to {e!r}")
  93         return None
  94
  95
  96 @mypyc_attr(patchable=True)
  97 def parse_pyproject_toml(path_config: str) -> Dict[str, Any]:
  98     """Parse a pyproject toml file, pulling out relevant parts for Black
  99
 100     If parsing fails, will raise a tomli.TOMLDecodeError
 101     """
 102     with open(path_config, "rb") as f:
 103         pyproject_toml = tomli.load(f)
 104     config = pyproject_toml.get("tool", {}).get("black", {})
 105     return {k.replace("--", "").replace("-", "_"): v for k, v in config.items()}
 106
 107
 108 @lru_cache()
 109 def find_user_pyproject_toml() -> Path:
 110     r"""Return the path to the top-level user configuration for black.
 111
 112     This looks for ~\.black on Windows and ~/.config/black on Linux and other
 113     Unix systems.
 114     """
 115     if sys.platform == "win32":
 116         # Windows
 117         user_config_path = Path.home() / ".black"
 118     else:
 119         config_root = os.environ.get("XDG_CONFIG_HOME", "~/.config")
 120         user_config_path = Path(config_root).expanduser() / "black"
 121     return user_config_path.resolve()
 122
 123
 124 @lru_cache()
 125 def get_gitignore(root: Path) -> PathSpec:
 126     """Return a PathSpec matching gitignore content if present."""
 127     gitignore = root / ".gitignore"
 128     lines: List[str] = []
 129     if gitignore.is_file():
 130         with gitignore.open(encoding="utf-8") as gf:
 131             lines = gf.readlines()
 132     try:
 133         return PathSpec.from_lines("gitwildmatch", lines)
 134     except GitWildMatchPatternError as e:
 135         err(f"Could not parse {gitignore}: {e}")
 136         raise
 137
 138
 139 def normalize_path_maybe_ignore(
 140     path: Path,
 141     root: Path,
 142     report: Optional[Report] = None,
 143 ) -> Optional[str]:
 144     """Normalize `path`. May return `None` if `path` was ignored.
 145
 146     `report` is where "path ignored" output goes.
 147     """
 148     try:
 149         abspath = path if path.is_absolute() else Path.cwd() / path
 150         normalized_path = abspath.resolve().relative_to(root).as_posix()
 151     except OSError as e:
 152         if report:
 153             report.path_ignored(path, f"cannot be read because {e}")
 154         return None
 155
 156     except ValueError:
 157         if path.is_symlink():
 158             if report:
 159                 report.path_ignored(
 160                     path, f"is a symbolic link that points outside {root}"
 161                 )
 162             return None
 163
 164         raise
 165
 166     return normalized_path
 167
 168
 169 def path_is_excluded(
 170     normalized_path: str,
 171     pattern: Optional[Pattern[str]],
 172 ) -> bool:
 173     match = pattern.search(normalized_path) if pattern else None
 174     return bool(match and match.group(0))
 175
 176
 177 def gen_python_files(
 178     paths: Iterable[Path],
 179     root: Path,
 180     include: Pattern[str],
 181     exclude: Pattern[str],
 182     extend_exclude: Optional[Pattern[str]],
 183     force_exclude: Optional[Pattern[str]],
 184     report: Report,
 185     gitignore: Optional[PathSpec],
 186     *,
 187     verbose: bool,
 188     quiet: bool,
 189 ) -> Iterator[Path]:
 190     """Generate all files under `path` whose paths are not excluded by the
 191     `exclude_regex`, `extend_exclude`, or `force_exclude` regexes,
 192     but are included by the `include` regex.
 193
 194     Symbolic links pointing outside of the `root` directory are ignored.
 195
 196     `report` is where output about exclusions goes.
 197     """
 198     assert root.is_absolute(), f"INTERNAL ERROR: `root` must be absolute but is {root}"
 199     for child in paths:
 200         normalized_path = normalize_path_maybe_ignore(child, root, report)
 201         if normalized_path is None:
 202             continue
 203
 204         # First ignore files matching .gitignore, if passed
 205         if gitignore is not None and gitignore.match_file(normalized_path):
 206             report.path_ignored(child, "matches the .gitignore file content")
 207             continue
 208
 209         # Then ignore with `--exclude` `--extend-exclude` and `--force-exclude` options.
 210         normalized_path = "/" + normalized_path
 211         if child.is_dir():
 212             normalized_path += "/"
 213
 214         if path_is_excluded(normalized_path, exclude):
 215             report.path_ignored(child, "matches the --exclude regular expression")
 216             continue
 217
 218         if path_is_excluded(normalized_path, extend_exclude):
 219             report.path_ignored(
 220                 child, "matches the --extend-exclude regular expression"
 221             )
 222             continue
 223
 224         if path_is_excluded(normalized_path, force_exclude):
 225             report.path_ignored(child, "matches the --force-exclude regular expression")
 226             continue
 227
 228         if child.is_dir():
 229             # If gitignore is None, gitignore usage is disabled, while a Falsey
 230             # gitignore is when the directory doesn't have a .gitignore file.
 231             yield from gen_python_files(
 232                 child.iterdir(),
 233                 root,
 234                 include,
 235                 exclude,
 236                 extend_exclude,
 237                 force_exclude,
 238                 report,
 239                 gitignore + get_gitignore(child) if gitignore is not None else None,
 240                 verbose=verbose,
 241                 quiet=quiet,
 242             )
 243
 244         elif child.is_file():
 245             if child.suffix == ".ipynb" and not jupyter_dependencies_are_installed(
 246                 verbose=verbose, quiet=quiet
 247             ):
 248                 continue
 249             include_match = include.search(normalized_path) if include else True
 250             if include_match:
 251                 yield child
 252
 253
 254 def wrap_stream_for_windows(
 255     f: io.TextIOWrapper,
 256 ) -> Union[io.TextIOWrapper, "colorama.AnsiToWin32"]:
 257     """
 258     Wrap stream with colorama's wrap_stream so colors are shown on Windows.
 259
 260     If `colorama` is unavailable, the original stream is returned unmodified.
 261     Otherwise, the `wrap_stream()` function determines whether the stream needs
 262     to be wrapped for a Windows environment and will accordingly either return
 263     an `AnsiToWin32` wrapper or the original stream.
 264     """
 265     try:
 266         from colorama.initialise import wrap_stream
 267     except ImportError:
 268         return f
 269     else:
 270         # Set `strip=False` to avoid needing to modify test_express_diff_with_color.
 271         return wrap_stream(f, convert=None, strip=False, autoreset=False, wrap=True)