Olivier Deprez | f4ef2d0 | 2021-04-20 13:36:24 +0200 | [diff] [blame] | 1 | """Common operations on Posix pathnames. |
| 2 | |
| 3 | Instead of importing this module directly, import os and refer to |
| 4 | this module as os.path. The "os.path" name is an alias for this |
| 5 | module on Posix systems; on other systems (e.g. Windows), |
| 6 | os.path provides the same operations in a manner specific to that |
| 7 | platform, and is an alias to another module (e.g. ntpath). |
| 8 | |
| 9 | Some of this can actually be useful on non-Posix systems too, e.g. |
| 10 | for manipulation of the pathname component of URLs. |
| 11 | """ |
| 12 | |
| 13 | # Strings representing various path-related bits and pieces. |
| 14 | # These are primarily for export; internally, they are hardcoded. |
| 15 | # Should be set before imports for resolving cyclic dependency. |
| 16 | curdir = '.' |
| 17 | pardir = '..' |
| 18 | extsep = '.' |
| 19 | sep = '/' |
| 20 | pathsep = ':' |
| 21 | defpath = '/bin:/usr/bin' |
| 22 | altsep = None |
| 23 | devnull = '/dev/null' |
| 24 | |
| 25 | import os |
| 26 | import sys |
| 27 | import stat |
| 28 | import genericpath |
| 29 | from genericpath import * |
| 30 | |
| 31 | __all__ = ["normcase","isabs","join","splitdrive","split","splitext", |
| 32 | "basename","dirname","commonprefix","getsize","getmtime", |
| 33 | "getatime","getctime","islink","exists","lexists","isdir","isfile", |
| 34 | "ismount", "expanduser","expandvars","normpath","abspath", |
| 35 | "samefile","sameopenfile","samestat", |
| 36 | "curdir","pardir","sep","pathsep","defpath","altsep","extsep", |
| 37 | "devnull","realpath","supports_unicode_filenames","relpath", |
| 38 | "commonpath"] |
| 39 | |
| 40 | |
| 41 | def _get_sep(path): |
| 42 | if isinstance(path, bytes): |
| 43 | return b'/' |
| 44 | else: |
| 45 | return '/' |
| 46 | |
| 47 | # Normalize the case of a pathname. Trivial in Posix, string.lower on Mac. |
| 48 | # On MS-DOS this may also turn slashes into backslashes; however, other |
| 49 | # normalizations (such as optimizing '../' away) are not allowed |
| 50 | # (another function should be defined to do that). |
| 51 | |
| 52 | def normcase(s): |
| 53 | """Normalize case of pathname. Has no effect under Posix""" |
| 54 | return os.fspath(s) |
| 55 | |
| 56 | |
| 57 | # Return whether a path is absolute. |
| 58 | # Trivial in Posix, harder on the Mac or MS-DOS. |
| 59 | |
| 60 | def isabs(s): |
| 61 | """Test whether a path is absolute""" |
| 62 | s = os.fspath(s) |
| 63 | sep = _get_sep(s) |
| 64 | return s.startswith(sep) |
| 65 | |
| 66 | |
| 67 | # Join pathnames. |
| 68 | # Ignore the previous parts if a part is absolute. |
| 69 | # Insert a '/' unless the first part is empty or already ends in '/'. |
| 70 | |
| 71 | def join(a, *p): |
| 72 | """Join two or more pathname components, inserting '/' as needed. |
| 73 | If any component is an absolute path, all previous path components |
| 74 | will be discarded. An empty last part will result in a path that |
| 75 | ends with a separator.""" |
| 76 | a = os.fspath(a) |
| 77 | sep = _get_sep(a) |
| 78 | path = a |
| 79 | try: |
| 80 | if not p: |
| 81 | path[:0] + sep #23780: Ensure compatible data type even if p is null. |
| 82 | for b in map(os.fspath, p): |
| 83 | if b.startswith(sep): |
| 84 | path = b |
| 85 | elif not path or path.endswith(sep): |
| 86 | path += b |
| 87 | else: |
| 88 | path += sep + b |
| 89 | except (TypeError, AttributeError, BytesWarning): |
| 90 | genericpath._check_arg_types('join', a, *p) |
| 91 | raise |
| 92 | return path |
| 93 | |
| 94 | |
| 95 | # Split a path in head (everything up to the last '/') and tail (the |
| 96 | # rest). If the path ends in '/', tail will be empty. If there is no |
| 97 | # '/' in the path, head will be empty. |
| 98 | # Trailing '/'es are stripped from head unless it is the root. |
| 99 | |
| 100 | def split(p): |
| 101 | """Split a pathname. Returns tuple "(head, tail)" where "tail" is |
| 102 | everything after the final slash. Either part may be empty.""" |
| 103 | p = os.fspath(p) |
| 104 | sep = _get_sep(p) |
| 105 | i = p.rfind(sep) + 1 |
| 106 | head, tail = p[:i], p[i:] |
| 107 | if head and head != sep*len(head): |
| 108 | head = head.rstrip(sep) |
| 109 | return head, tail |
| 110 | |
| 111 | |
| 112 | # Split a path in root and extension. |
| 113 | # The extension is everything starting at the last dot in the last |
| 114 | # pathname component; the root is everything before that. |
| 115 | # It is always true that root + ext == p. |
| 116 | |
| 117 | def splitext(p): |
| 118 | p = os.fspath(p) |
| 119 | if isinstance(p, bytes): |
| 120 | sep = b'/' |
| 121 | extsep = b'.' |
| 122 | else: |
| 123 | sep = '/' |
| 124 | extsep = '.' |
| 125 | return genericpath._splitext(p, sep, None, extsep) |
| 126 | splitext.__doc__ = genericpath._splitext.__doc__ |
| 127 | |
| 128 | # Split a pathname into a drive specification and the rest of the |
| 129 | # path. Useful on DOS/Windows/NT; on Unix, the drive is always empty. |
| 130 | |
| 131 | def splitdrive(p): |
| 132 | """Split a pathname into drive and path. On Posix, drive is always |
| 133 | empty.""" |
| 134 | p = os.fspath(p) |
| 135 | return p[:0], p |
| 136 | |
| 137 | |
| 138 | # Return the tail (basename) part of a path, same as split(path)[1]. |
| 139 | |
| 140 | def basename(p): |
| 141 | """Returns the final component of a pathname""" |
| 142 | p = os.fspath(p) |
| 143 | sep = _get_sep(p) |
| 144 | i = p.rfind(sep) + 1 |
| 145 | return p[i:] |
| 146 | |
| 147 | |
| 148 | # Return the head (dirname) part of a path, same as split(path)[0]. |
| 149 | |
| 150 | def dirname(p): |
| 151 | """Returns the directory component of a pathname""" |
| 152 | p = os.fspath(p) |
| 153 | sep = _get_sep(p) |
| 154 | i = p.rfind(sep) + 1 |
| 155 | head = p[:i] |
| 156 | if head and head != sep*len(head): |
| 157 | head = head.rstrip(sep) |
| 158 | return head |
| 159 | |
| 160 | |
| 161 | # Is a path a symbolic link? |
| 162 | # This will always return false on systems where os.lstat doesn't exist. |
| 163 | |
| 164 | def islink(path): |
| 165 | """Test whether a path is a symbolic link""" |
| 166 | try: |
| 167 | st = os.lstat(path) |
| 168 | except (OSError, ValueError, AttributeError): |
| 169 | return False |
| 170 | return stat.S_ISLNK(st.st_mode) |
| 171 | |
| 172 | # Being true for dangling symbolic links is also useful. |
| 173 | |
| 174 | def lexists(path): |
| 175 | """Test whether a path exists. Returns True for broken symbolic links""" |
| 176 | try: |
| 177 | os.lstat(path) |
| 178 | except (OSError, ValueError): |
| 179 | return False |
| 180 | return True |
| 181 | |
| 182 | |
| 183 | # Is a path a mount point? |
| 184 | # (Does this work for all UNIXes? Is it even guaranteed to work by Posix?) |
| 185 | |
| 186 | def ismount(path): |
| 187 | """Test whether a path is a mount point""" |
| 188 | try: |
| 189 | s1 = os.lstat(path) |
| 190 | except (OSError, ValueError): |
| 191 | # It doesn't exist -- so not a mount point. :-) |
| 192 | return False |
| 193 | else: |
| 194 | # A symlink can never be a mount point |
| 195 | if stat.S_ISLNK(s1.st_mode): |
| 196 | return False |
| 197 | |
| 198 | if isinstance(path, bytes): |
| 199 | parent = join(path, b'..') |
| 200 | else: |
| 201 | parent = join(path, '..') |
| 202 | parent = realpath(parent) |
| 203 | try: |
| 204 | s2 = os.lstat(parent) |
| 205 | except (OSError, ValueError): |
| 206 | return False |
| 207 | |
| 208 | dev1 = s1.st_dev |
| 209 | dev2 = s2.st_dev |
| 210 | if dev1 != dev2: |
| 211 | return True # path/.. on a different device as path |
| 212 | ino1 = s1.st_ino |
| 213 | ino2 = s2.st_ino |
| 214 | if ino1 == ino2: |
| 215 | return True # path/.. is the same i-node as path |
| 216 | return False |
| 217 | |
| 218 | |
| 219 | # Expand paths beginning with '~' or '~user'. |
| 220 | # '~' means $HOME; '~user' means that user's home directory. |
| 221 | # If the path doesn't begin with '~', or if the user or $HOME is unknown, |
| 222 | # the path is returned unchanged (leaving error reporting to whatever |
| 223 | # function is called with the expanded path as argument). |
| 224 | # See also module 'glob' for expansion of *, ? and [...] in pathnames. |
| 225 | # (A function should also be defined to do full *sh-style environment |
| 226 | # variable expansion.) |
| 227 | |
| 228 | def expanduser(path): |
| 229 | """Expand ~ and ~user constructions. If user or $HOME is unknown, |
| 230 | do nothing.""" |
| 231 | path = os.fspath(path) |
| 232 | if isinstance(path, bytes): |
| 233 | tilde = b'~' |
| 234 | else: |
| 235 | tilde = '~' |
| 236 | if not path.startswith(tilde): |
| 237 | return path |
| 238 | sep = _get_sep(path) |
| 239 | i = path.find(sep, 1) |
| 240 | if i < 0: |
| 241 | i = len(path) |
| 242 | if i == 1: |
| 243 | if 'HOME' not in os.environ: |
| 244 | import pwd |
| 245 | try: |
| 246 | userhome = pwd.getpwuid(os.getuid()).pw_dir |
| 247 | except KeyError: |
| 248 | # bpo-10496: if the current user identifier doesn't exist in the |
| 249 | # password database, return the path unchanged |
| 250 | return path |
| 251 | else: |
| 252 | userhome = os.environ['HOME'] |
| 253 | else: |
| 254 | import pwd |
| 255 | name = path[1:i] |
| 256 | if isinstance(name, bytes): |
| 257 | name = str(name, 'ASCII') |
| 258 | try: |
| 259 | pwent = pwd.getpwnam(name) |
| 260 | except KeyError: |
| 261 | # bpo-10496: if the user name from the path doesn't exist in the |
| 262 | # password database, return the path unchanged |
| 263 | return path |
| 264 | userhome = pwent.pw_dir |
| 265 | if isinstance(path, bytes): |
| 266 | userhome = os.fsencode(userhome) |
| 267 | root = b'/' |
| 268 | else: |
| 269 | root = '/' |
| 270 | userhome = userhome.rstrip(root) |
| 271 | return (userhome + path[i:]) or root |
| 272 | |
| 273 | |
| 274 | # Expand paths containing shell variable substitutions. |
| 275 | # This expands the forms $variable and ${variable} only. |
| 276 | # Non-existent variables are left unchanged. |
| 277 | |
| 278 | _varprog = None |
| 279 | _varprogb = None |
| 280 | |
| 281 | def expandvars(path): |
| 282 | """Expand shell variables of form $var and ${var}. Unknown variables |
| 283 | are left unchanged.""" |
| 284 | path = os.fspath(path) |
| 285 | global _varprog, _varprogb |
| 286 | if isinstance(path, bytes): |
| 287 | if b'$' not in path: |
| 288 | return path |
| 289 | if not _varprogb: |
| 290 | import re |
| 291 | _varprogb = re.compile(br'\$(\w+|\{[^}]*\})', re.ASCII) |
| 292 | search = _varprogb.search |
| 293 | start = b'{' |
| 294 | end = b'}' |
| 295 | environ = getattr(os, 'environb', None) |
| 296 | else: |
| 297 | if '$' not in path: |
| 298 | return path |
| 299 | if not _varprog: |
| 300 | import re |
| 301 | _varprog = re.compile(r'\$(\w+|\{[^}]*\})', re.ASCII) |
| 302 | search = _varprog.search |
| 303 | start = '{' |
| 304 | end = '}' |
| 305 | environ = os.environ |
| 306 | i = 0 |
| 307 | while True: |
| 308 | m = search(path, i) |
| 309 | if not m: |
| 310 | break |
| 311 | i, j = m.span(0) |
| 312 | name = m.group(1) |
| 313 | if name.startswith(start) and name.endswith(end): |
| 314 | name = name[1:-1] |
| 315 | try: |
| 316 | if environ is None: |
| 317 | value = os.fsencode(os.environ[os.fsdecode(name)]) |
| 318 | else: |
| 319 | value = environ[name] |
| 320 | except KeyError: |
| 321 | i = j |
| 322 | else: |
| 323 | tail = path[j:] |
| 324 | path = path[:i] + value |
| 325 | i = len(path) |
| 326 | path += tail |
| 327 | return path |
| 328 | |
| 329 | |
| 330 | # Normalize a path, e.g. A//B, A/./B and A/foo/../B all become A/B. |
| 331 | # It should be understood that this may change the meaning of the path |
| 332 | # if it contains symbolic links! |
| 333 | |
| 334 | def normpath(path): |
| 335 | """Normalize path, eliminating double slashes, etc.""" |
| 336 | path = os.fspath(path) |
| 337 | if isinstance(path, bytes): |
| 338 | sep = b'/' |
| 339 | empty = b'' |
| 340 | dot = b'.' |
| 341 | dotdot = b'..' |
| 342 | else: |
| 343 | sep = '/' |
| 344 | empty = '' |
| 345 | dot = '.' |
| 346 | dotdot = '..' |
| 347 | if path == empty: |
| 348 | return dot |
| 349 | initial_slashes = path.startswith(sep) |
| 350 | # POSIX allows one or two initial slashes, but treats three or more |
| 351 | # as single slash. |
| 352 | if (initial_slashes and |
| 353 | path.startswith(sep*2) and not path.startswith(sep*3)): |
| 354 | initial_slashes = 2 |
| 355 | comps = path.split(sep) |
| 356 | new_comps = [] |
| 357 | for comp in comps: |
| 358 | if comp in (empty, dot): |
| 359 | continue |
| 360 | if (comp != dotdot or (not initial_slashes and not new_comps) or |
| 361 | (new_comps and new_comps[-1] == dotdot)): |
| 362 | new_comps.append(comp) |
| 363 | elif new_comps: |
| 364 | new_comps.pop() |
| 365 | comps = new_comps |
| 366 | path = sep.join(comps) |
| 367 | if initial_slashes: |
| 368 | path = sep*initial_slashes + path |
| 369 | return path or dot |
| 370 | |
| 371 | |
| 372 | def abspath(path): |
| 373 | """Return an absolute path.""" |
| 374 | path = os.fspath(path) |
| 375 | if not isabs(path): |
| 376 | if isinstance(path, bytes): |
| 377 | cwd = os.getcwdb() |
| 378 | else: |
| 379 | cwd = os.getcwd() |
| 380 | path = join(cwd, path) |
| 381 | return normpath(path) |
| 382 | |
| 383 | |
| 384 | # Return a canonical path (i.e. the absolute location of a file on the |
| 385 | # filesystem). |
| 386 | |
| 387 | def realpath(filename): |
| 388 | """Return the canonical path of the specified filename, eliminating any |
| 389 | symbolic links encountered in the path.""" |
| 390 | filename = os.fspath(filename) |
| 391 | path, ok = _joinrealpath(filename[:0], filename, {}) |
| 392 | return abspath(path) |
| 393 | |
| 394 | # Join two paths, normalizing and eliminating any symbolic links |
| 395 | # encountered in the second path. |
| 396 | def _joinrealpath(path, rest, seen): |
| 397 | if isinstance(path, bytes): |
| 398 | sep = b'/' |
| 399 | curdir = b'.' |
| 400 | pardir = b'..' |
| 401 | else: |
| 402 | sep = '/' |
| 403 | curdir = '.' |
| 404 | pardir = '..' |
| 405 | |
| 406 | if isabs(rest): |
| 407 | rest = rest[1:] |
| 408 | path = sep |
| 409 | |
| 410 | while rest: |
| 411 | name, _, rest = rest.partition(sep) |
| 412 | if not name or name == curdir: |
| 413 | # current dir |
| 414 | continue |
| 415 | if name == pardir: |
| 416 | # parent dir |
| 417 | if path: |
| 418 | path, name = split(path) |
| 419 | if name == pardir: |
| 420 | path = join(path, pardir, pardir) |
| 421 | else: |
| 422 | path = pardir |
| 423 | continue |
| 424 | newpath = join(path, name) |
| 425 | if not islink(newpath): |
| 426 | path = newpath |
| 427 | continue |
| 428 | # Resolve the symbolic link |
| 429 | if newpath in seen: |
| 430 | # Already seen this path |
| 431 | path = seen[newpath] |
| 432 | if path is not None: |
| 433 | # use cached value |
| 434 | continue |
| 435 | # The symlink is not resolved, so we must have a symlink loop. |
| 436 | # Return already resolved part + rest of the path unchanged. |
| 437 | return join(newpath, rest), False |
| 438 | seen[newpath] = None # not resolved symlink |
| 439 | path, ok = _joinrealpath(path, os.readlink(newpath), seen) |
| 440 | if not ok: |
| 441 | return join(path, rest), False |
| 442 | seen[newpath] = path # resolved symlink |
| 443 | |
| 444 | return path, True |
| 445 | |
| 446 | |
| 447 | supports_unicode_filenames = (sys.platform == 'darwin') |
| 448 | |
| 449 | def relpath(path, start=None): |
| 450 | """Return a relative version of a path""" |
| 451 | |
| 452 | if not path: |
| 453 | raise ValueError("no path specified") |
| 454 | |
| 455 | path = os.fspath(path) |
| 456 | if isinstance(path, bytes): |
| 457 | curdir = b'.' |
| 458 | sep = b'/' |
| 459 | pardir = b'..' |
| 460 | else: |
| 461 | curdir = '.' |
| 462 | sep = '/' |
| 463 | pardir = '..' |
| 464 | |
| 465 | if start is None: |
| 466 | start = curdir |
| 467 | else: |
| 468 | start = os.fspath(start) |
| 469 | |
| 470 | try: |
| 471 | start_list = [x for x in abspath(start).split(sep) if x] |
| 472 | path_list = [x for x in abspath(path).split(sep) if x] |
| 473 | # Work out how much of the filepath is shared by start and path. |
| 474 | i = len(commonprefix([start_list, path_list])) |
| 475 | |
| 476 | rel_list = [pardir] * (len(start_list)-i) + path_list[i:] |
| 477 | if not rel_list: |
| 478 | return curdir |
| 479 | return join(*rel_list) |
| 480 | except (TypeError, AttributeError, BytesWarning, DeprecationWarning): |
| 481 | genericpath._check_arg_types('relpath', path, start) |
| 482 | raise |
| 483 | |
| 484 | |
| 485 | # Return the longest common sub-path of the sequence of paths given as input. |
| 486 | # The paths are not normalized before comparing them (this is the |
| 487 | # responsibility of the caller). Any trailing separator is stripped from the |
| 488 | # returned path. |
| 489 | |
| 490 | def commonpath(paths): |
| 491 | """Given a sequence of path names, returns the longest common sub-path.""" |
| 492 | |
| 493 | if not paths: |
| 494 | raise ValueError('commonpath() arg is an empty sequence') |
| 495 | |
| 496 | paths = tuple(map(os.fspath, paths)) |
| 497 | if isinstance(paths[0], bytes): |
| 498 | sep = b'/' |
| 499 | curdir = b'.' |
| 500 | else: |
| 501 | sep = '/' |
| 502 | curdir = '.' |
| 503 | |
| 504 | try: |
| 505 | split_paths = [path.split(sep) for path in paths] |
| 506 | |
| 507 | try: |
| 508 | isabs, = set(p[:1] == sep for p in paths) |
| 509 | except ValueError: |
| 510 | raise ValueError("Can't mix absolute and relative paths") from None |
| 511 | |
| 512 | split_paths = [[c for c in s if c and c != curdir] for s in split_paths] |
| 513 | s1 = min(split_paths) |
| 514 | s2 = max(split_paths) |
| 515 | common = s1 |
| 516 | for i, c in enumerate(s1): |
| 517 | if c != s2[i]: |
| 518 | common = s1[:i] |
| 519 | break |
| 520 | |
| 521 | prefix = sep if isabs else sep[:0] |
| 522 | return prefix + sep.join(common) |
| 523 | except (TypeError, AttributeError): |
| 524 | genericpath._check_arg_types('commonpath', *paths) |
| 525 | raise |