| 1 | import re |
| 2 | import unicodedata |
| 3 | from pathlib import Path |
| 4 | from typing import Final, Optional |
| 5 | |
| 6 | # Forbidden characters: |
| 7 | # Linux/Unix: / and NULL byte |
| 8 | # Windows: < > : " / \ | ? * and ASCII control characters (0-31) |
| 9 | # Shell-sensitive: ~ to prevent accidental home directory access |
| 10 | FORBIDDEN_CHARS_RE: Final = re.compile(r'[<>:"|?*~/\\\x00-\x1f\x7f]') |
| 11 | |
| 12 | # Windows reserved filenames |
| 13 | WINDOWS_RESERVED: Final = frozenset({ |
| 14 | "CON", "PRN", "AUX", "NUL", "CONIN$", "CONOUT$", |
| 15 | "COM1", "COM2", "COM3", "COM4", "COM5", "COM6", "COM7", "COM8", "COM9", |
| 16 | "LPT1", "LPT2", "LPT3", "LPT4", "LPT5", "LPT6", "LPT7", "LPT8", "LPT9" |
| 17 | }) |
| 18 | |
| 19 | FILENAME_MAX_LENGTH: Final = 255 |
| 20 | |
| 21 | def safe_filename(filename: str) -> Optional[str]: |
| 22 | # Normalize Unicode (NFC) |
| 23 | filename = unicodedata.normalize("NFC", str(filename)) |
| 24 | # Replace forbidden chars |
| 25 | filename = FORBIDDEN_CHARS_RE.sub("_", filename) |
| 26 | # Remove leading/trailing spaces and trailing dots |
| 27 | filename = filename.lstrip(" ").rstrip(". ") |
| 28 | |
| 29 | path = Path(filename) |
| 30 | suffixes = ''.join(path.suffixes) |
| 31 | stem = path.name[:-len(suffixes)] if suffixes else path.name |
| 32 | |
| 33 | # Check Windows reserved names |
| 34 | if stem.upper() in WINDOWS_RESERVED: |
| 35 | filename = f"{stem}-{suffixes}" |
| 36 | |
| 37 | # Truncate if too long |
| 38 | if len(filename) > FILENAME_MAX_LENGTH: |
| 39 | max_stem_len = FILENAME_MAX_LENGTH - len(suffixes) |
| 40 | if max_stem_len > 0: |
| 41 | # Truncate filename |
| 42 | stem = stem[:max_stem_len] |
| 43 | filename = stem + suffixes |
| 44 | else: |
| 45 | # Extension is too long, truncate everything |
| 46 | filename = filename[:FILENAME_MAX_LENGTH] |
| 47 | if not filename: |
| 48 | return None |
| 49 | return filename |