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, RuntimeError) 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     May raise:
 116     - RuntimeError: if the current user has no homedir
 117     - PermissionError: if the current process cannot access the user's homedir
 118     """
 119     if sys.platform == "win32":
 120         # Windows
 121         user_config_path = Path.home() / ".black"
 122     else:
 123         config_root = os.environ.get("XDG_CONFIG_HOME", "~/.config")
 124         user_config_path = Path(config_root).expanduser() / "black"
 125     return user_config_path.resolve()
 126
 127
 128 @lru_cache()
 129 def get_gitignore(root: Path) -> PathSpec:
 130     """Return a PathSpec matching gitignore content if present."""
 131     gitignore = root / ".gitignore"
 132     lines: List[str] = []
 133     if gitignore.is_file():
 134         with gitignore.open(encoding="utf-8") as gf:
 135             lines = gf.readlines()
 136     try:
 137         return PathSpec.from_lines("gitwildmatch", lines)
 138     except GitWildMatchPatternError as e:
 139         err(f"Could not parse {gitignore}: {e}")
 140         raise
 141
 142
 143 def normalize_path_maybe_ignore(
 144     path: Path,
 145     root: Path,
 146     report: Optional[Report] = None,
 147 ) -> Optional[str]:
 148     """Normalize `path`. May return `None` if `path` was ignored.
 149
 150     `report` is where "path ignored" output goes.
 151     """
 152     try:
 153         abspath = path if path.is_absolute() else Path.cwd() / path
 154         normalized_path = abspath.resolve().relative_to(root).as_posix()
 155     except OSError as e:
 156         if report:
 157             report.path_ignored(path, f"cannot be read because {e}")
 158         return None
 159
 160     except ValueError:
 161         if path.is_symlink():
 162             if report:
 163                 report.path_ignored(
 164                     path, f"is a symbolic link that points outside {root}"
 165                 )
 166             return None
 167
 168         raise
 169
 170     return normalized_path
 171
 172
 173 def path_is_excluded(
 174     normalized_path: str,
 175     pattern: Optional[Pattern[str]],
 176 ) -> bool:
 177     match = pattern.search(normalized_path) if pattern else None
 178     return bool(match and match.group(0))
 179
 180
 181 def gen_python_files(
 182     paths: Iterable[Path],
 183     root: Path,
 184     include: Pattern[str],
 185     exclude: Pattern[str],
 186     extend_exclude: Optional[Pattern[str]],
 187     force_exclude: Optional[Pattern[str]],
 188     report: Report,
 189     gitignore: Optional[PathSpec],
 190     *,
 191     verbose: bool,
 192     quiet: bool,
 193 ) -> Iterator[Path]:
 194     """Generate all files under `path` whose paths are not excluded by the
 195     `exclude_regex`, `extend_exclude`, or `force_exclude` regexes,
 196     but are included by the `include` regex.
 197
 198     Symbolic links pointing outside of the `root` directory are ignored.
 199
 200     `report` is where output about exclusions goes.
 201     """
 202     assert root.is_absolute(), f"INTERNAL ERROR: `root` must be absolute but is {root}"
 203     for child in paths:
 204         normalized_path = normalize_path_maybe_ignore(child, root, report)
 205         if normalized_path is None:
 206             continue
 207
 208         # First ignore files matching .gitignore, if passed
 209         if gitignore is not None and gitignore.match_file(normalized_path):
 210             report.path_ignored(child, "matches the .gitignore file content")
 211             continue
 212
 213         # Then ignore with `--exclude` `--extend-exclude` and `--force-exclude` options.
 214         normalized_path = "/" + normalized_path
 215         if child.is_dir():
 216             normalized_path += "/"
 217
 218         if path_is_excluded(normalized_path, exclude):
 219             report.path_ignored(child, "matches the --exclude regular expression")
 220             continue
 221
 222         if path_is_excluded(normalized_path, extend_exclude):
 223             report.path_ignored(
 224                 child, "matches the --extend-exclude regular expression"
 225             )
 226             continue
 227
 228         if path_is_excluded(normalized_path, force_exclude):
 229             report.path_ignored(child, "matches the --force-exclude regular expression")
 230             continue
 231
 232         if child.is_dir():
 233             # If gitignore is None, gitignore usage is disabled, while a Falsey
 234             # gitignore is when the directory doesn't have a .gitignore file.
 235             yield from gen_python_files(
 236                 child.iterdir(),
 237                 root,
 238                 include,
 239                 exclude,
 240                 extend_exclude,
 241                 force_exclude,
 242                 report,
 243                 gitignore + get_gitignore(child) if gitignore is not None else None,
 244                 verbose=verbose,
 245                 quiet=quiet,
 246             )
 247
 248         elif child.is_file():
 249             if child.suffix == ".ipynb" and not jupyter_dependencies_are_installed(
 250                 verbose=verbose, quiet=quiet
 251             ):
 252                 continue
 253             include_match = include.search(normalized_path) if include else True
 254             if include_match:
 255                 yield child
 256
 257
 258 def wrap_stream_for_windows(
 259     f: io.TextIOWrapper,
 260 ) -> Union[io.TextIOWrapper, "colorama.AnsiToWin32"]:
 261     """
 262     Wrap stream with colorama's wrap_stream so colors are shown on Windows.
 263
 264     If `colorama` is unavailable, the original stream is returned unmodified.
 265     Otherwise, the `wrap_stream()` function determines whether the stream needs
 266     to be wrapped for a Windows environment and will accordingly either return
 267     an `AnsiToWin32` wrapper or the original stream.
 268     """
 269     try:
 270         from colorama.initialise import wrap_stream
 271     except ImportError:
 272         return f
 273     else:
 274         # Set `strip=False` to avoid needing to modify test_express_diff_with_color.
 275         return wrap_stream(f, convert=None, strip=False, autoreset=False, wrap=True)