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