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 pathspec import PathSpec
  21 import tomli
  22
  23 from black.output import err
  24 from black.report import Report
  25 from black.handle_ipynb_magics import jupyter_dependencies_are_installed
  26
  27 if TYPE_CHECKING:
  28     import colorama  # noqa: F401
  29
  30
  31 @lru_cache()
  32 def find_project_root(srcs: Sequence[str]) -> Path:
  33     """Return a directory containing .git, .hg, or pyproject.toml.
  34
  35     That directory will be a common parent of all files and directories
  36     passed in `srcs`.
  37
  38     If no directory in the tree contains a marker that would specify it's the
  39     project root, the root of the file system is returned.
  40     """
  41     if not srcs:
  42         srcs = [str(Path.cwd().resolve())]
  43
  44     path_srcs = [Path(Path.cwd(), src).resolve() for src in srcs]
  45
  46     # A list of lists of parents for each 'src'. 'src' is included as a
  47     # "parent" of itself if it is a directory
  48     src_parents = [
  49         list(path.parents) + ([path] if path.is_dir() else []) for path in path_srcs
  50     ]
  51
  52     common_base = max(
  53         set.intersection(*(set(parents) for parents in src_parents)),
  54         key=lambda path: path.parts,
  55     )
  56
  57     for directory in (common_base, *common_base.parents):
  58         if (directory / ".git").exists():
  59             return directory
  60
  61         if (directory / ".hg").is_dir():
  62             return directory
  63
  64         if (directory / "pyproject.toml").is_file():
  65             return directory
  66
  67     return directory
  68
  69
  70 def find_pyproject_toml(path_search_start: Tuple[str, ...]) -> Optional[str]:
  71     """Find the absolute filepath to a pyproject.toml if it exists"""
  72     path_project_root = find_project_root(path_search_start)
  73     path_pyproject_toml = path_project_root / "pyproject.toml"
  74     if path_pyproject_toml.is_file():
  75         return str(path_pyproject_toml)
  76
  77     try:
  78         path_user_pyproject_toml = find_user_pyproject_toml()
  79         return (
  80             str(path_user_pyproject_toml)
  81             if path_user_pyproject_toml.is_file()
  82             else None
  83         )
  84     except PermissionError as e:
  85         # We do not have access to the user-level config directory, so ignore it.
  86         err(f"Ignoring user configuration directory due to {e!r}")
  87         return None
  88
  89
  90 def parse_pyproject_toml(path_config: str) -> Dict[str, Any]:
  91     """Parse a pyproject toml file, pulling out relevant parts for Black
  92
  93     If parsing fails, will raise a tomli.TOMLDecodeError
  94     """
  95     with open(path_config, encoding="utf8") as f:
  96         pyproject_toml = tomli.load(f)  # type: ignore  # due to deprecated API usage
  97     config = pyproject_toml.get("tool", {}).get("black", {})
  98     return {k.replace("--", "").replace("-", "_"): v for k, v in config.items()}
  99
 100
 101 @lru_cache()
 102 def find_user_pyproject_toml() -> Path:
 103     r"""Return the path to the top-level user configuration for black.
 104
 105     This looks for ~\.black on Windows and ~/.config/black on Linux and other
 106     Unix systems.
 107     """
 108     if sys.platform == "win32":
 109         # Windows
 110         user_config_path = Path.home() / ".black"
 111     else:
 112         config_root = os.environ.get("XDG_CONFIG_HOME", "~/.config")
 113         user_config_path = Path(config_root).expanduser() / "black"
 114     return user_config_path.resolve()
 115
 116
 117 @lru_cache()
 118 def get_gitignore(root: Path) -> PathSpec:
 119     """Return a PathSpec matching gitignore content if present."""
 120     gitignore = root / ".gitignore"
 121     lines: List[str] = []
 122     if gitignore.is_file():
 123         with gitignore.open(encoding="utf-8") as gf:
 124             lines = gf.readlines()
 125     return PathSpec.from_lines("gitwildmatch", lines)
 126
 127
 128 def normalize_path_maybe_ignore(
 129     path: Path, root: Path, report: Report
 130 ) -> Optional[str]:
 131     """Normalize `path`. May return `None` if `path` was ignored.
 132
 133     `report` is where "path ignored" output goes.
 134     """
 135     try:
 136         abspath = path if path.is_absolute() else Path.cwd() / path
 137         normalized_path = abspath.resolve().relative_to(root).as_posix()
 138     except OSError as e:
 139         report.path_ignored(path, f"cannot be read because {e}")
 140         return None
 141
 142     except ValueError:
 143         if path.is_symlink():
 144             report.path_ignored(path, f"is a symbolic link that points outside {root}")
 145             return None
 146
 147         raise
 148
 149     return normalized_path
 150
 151
 152 def path_is_excluded(
 153     normalized_path: str,
 154     pattern: Optional[Pattern[str]],
 155 ) -> bool:
 156     match = pattern.search(normalized_path) if pattern else None
 157     return bool(match and match.group(0))
 158
 159
 160 def gen_python_files(
 161     paths: Iterable[Path],
 162     root: Path,
 163     include: Pattern[str],
 164     exclude: Pattern[str],
 165     extend_exclude: Optional[Pattern[str]],
 166     force_exclude: Optional[Pattern[str]],
 167     report: Report,
 168     gitignore: Optional[PathSpec],
 169     *,
 170     verbose: bool,
 171     quiet: bool,
 172 ) -> Iterator[Path]:
 173     """Generate all files under `path` whose paths are not excluded by the
 174     `exclude_regex`, `extend_exclude`, or `force_exclude` regexes,
 175     but are included by the `include` regex.
 176
 177     Symbolic links pointing outside of the `root` directory are ignored.
 178
 179     `report` is where output about exclusions goes.
 180     """
 181     assert root.is_absolute(), f"INTERNAL ERROR: `root` must be absolute but is {root}"
 182     for child in paths:
 183         normalized_path = normalize_path_maybe_ignore(child, root, report)
 184         if normalized_path is None:
 185             continue
 186
 187         # First ignore files matching .gitignore, if passed
 188         if gitignore is not None and gitignore.match_file(normalized_path):
 189             report.path_ignored(child, "matches the .gitignore file content")
 190             continue
 191
 192         # Then ignore with `--exclude` `--extend-exclude` and `--force-exclude` options.
 193         normalized_path = "/" + normalized_path
 194         if child.is_dir():
 195             normalized_path += "/"
 196
 197         if path_is_excluded(normalized_path, exclude):
 198             report.path_ignored(child, "matches the --exclude regular expression")
 199             continue
 200
 201         if path_is_excluded(normalized_path, extend_exclude):
 202             report.path_ignored(
 203                 child, "matches the --extend-exclude regular expression"
 204             )
 205             continue
 206
 207         if path_is_excluded(normalized_path, force_exclude):
 208             report.path_ignored(child, "matches the --force-exclude regular expression")
 209             continue
 210
 211         if child.is_dir():
 212             # If gitignore is None, gitignore usage is disabled, while a Falsey
 213             # gitignore is when the directory doesn't have a .gitignore file.
 214             yield from gen_python_files(
 215                 child.iterdir(),
 216                 root,
 217                 include,
 218                 exclude,
 219                 extend_exclude,
 220                 force_exclude,
 221                 report,
 222                 gitignore + get_gitignore(child) if gitignore is not None else None,
 223                 verbose=verbose,
 224                 quiet=quiet,
 225             )
 226
 227         elif child.is_file():
 228             if child.suffix == ".ipynb" and not jupyter_dependencies_are_installed(
 229                 verbose=verbose, quiet=quiet
 230             ):
 231                 continue
 232             include_match = include.search(normalized_path) if include else True
 233             if include_match:
 234                 yield child
 235
 236
 237 def wrap_stream_for_windows(
 238     f: io.TextIOWrapper,
 239 ) -> Union[io.TextIOWrapper, "colorama.AnsiToWin32"]:
 240     """
 241     Wrap stream with colorama's wrap_stream so colors are shown on Windows.
 242
 243     If `colorama` is unavailable, the original stream is returned unmodified.
 244     Otherwise, the `wrap_stream()` function determines whether the stream needs
 245     to be wrapped for a Windows environment and will accordingly either return
 246     an `AnsiToWin32` wrapper or the original stream.
 247     """
 248     try:
 249         from colorama.initialise import wrap_stream
 250     except ImportError:
 251         return f
 252     else:
 253         # Set `strip=False` to avoid needing to modify test_express_diff_with_color.
 254         return wrap_stream(f, convert=None, strip=False, autoreset=False, wrap=True)