| 1 | n/a | """Filename globbing utility.""" |
|---|
| 2 | n/a | |
|---|
| 3 | n/a | import os |
|---|
| 4 | n/a | import re |
|---|
| 5 | n/a | import fnmatch |
|---|
| 6 | n/a | |
|---|
| 7 | n/a | __all__ = ["glob", "iglob", "escape"] |
|---|
| 8 | n/a | |
|---|
| 9 | n/a | def glob(pathname, *, recursive=False): |
|---|
| 10 | n/a | """Return a list of paths matching a pathname pattern. |
|---|
| 11 | n/a | |
|---|
| 12 | n/a | The pattern may contain simple shell-style wildcards a la |
|---|
| 13 | n/a | fnmatch. However, unlike fnmatch, filenames starting with a |
|---|
| 14 | n/a | dot are special cases that are not matched by '*' and '?' |
|---|
| 15 | n/a | patterns. |
|---|
| 16 | n/a | |
|---|
| 17 | n/a | If recursive is true, the pattern '**' will match any files and |
|---|
| 18 | n/a | zero or more directories and subdirectories. |
|---|
| 19 | n/a | """ |
|---|
| 20 | n/a | return list(iglob(pathname, recursive=recursive)) |
|---|
| 21 | n/a | |
|---|
| 22 | n/a | def iglob(pathname, *, recursive=False): |
|---|
| 23 | n/a | """Return an iterator which yields the paths matching a pathname pattern. |
|---|
| 24 | n/a | |
|---|
| 25 | n/a | The pattern may contain simple shell-style wildcards a la |
|---|
| 26 | n/a | fnmatch. However, unlike fnmatch, filenames starting with a |
|---|
| 27 | n/a | dot are special cases that are not matched by '*' and '?' |
|---|
| 28 | n/a | patterns. |
|---|
| 29 | n/a | |
|---|
| 30 | n/a | If recursive is true, the pattern '**' will match any files and |
|---|
| 31 | n/a | zero or more directories and subdirectories. |
|---|
| 32 | n/a | """ |
|---|
| 33 | n/a | it = _iglob(pathname, recursive, False) |
|---|
| 34 | n/a | if recursive and _isrecursive(pathname): |
|---|
| 35 | n/a | s = next(it) # skip empty string |
|---|
| 36 | n/a | assert not s |
|---|
| 37 | n/a | return it |
|---|
| 38 | n/a | |
|---|
| 39 | n/a | def _iglob(pathname, recursive, dironly): |
|---|
| 40 | n/a | dirname, basename = os.path.split(pathname) |
|---|
| 41 | n/a | if not has_magic(pathname): |
|---|
| 42 | n/a | assert not dironly |
|---|
| 43 | n/a | if basename: |
|---|
| 44 | n/a | if os.path.lexists(pathname): |
|---|
| 45 | n/a | yield pathname |
|---|
| 46 | n/a | else: |
|---|
| 47 | n/a | # Patterns ending with a slash should match only directories |
|---|
| 48 | n/a | if os.path.isdir(dirname): |
|---|
| 49 | n/a | yield pathname |
|---|
| 50 | n/a | return |
|---|
| 51 | n/a | if not dirname: |
|---|
| 52 | n/a | if recursive and _isrecursive(basename): |
|---|
| 53 | n/a | yield from _glob2(dirname, basename, dironly) |
|---|
| 54 | n/a | else: |
|---|
| 55 | n/a | yield from _glob1(dirname, basename, dironly) |
|---|
| 56 | n/a | return |
|---|
| 57 | n/a | # `os.path.split()` returns the argument itself as a dirname if it is a |
|---|
| 58 | n/a | # drive or UNC path. Prevent an infinite recursion if a drive or UNC path |
|---|
| 59 | n/a | # contains magic characters (i.e. r'\\?\C:'). |
|---|
| 60 | n/a | if dirname != pathname and has_magic(dirname): |
|---|
| 61 | n/a | dirs = _iglob(dirname, recursive, True) |
|---|
| 62 | n/a | else: |
|---|
| 63 | n/a | dirs = [dirname] |
|---|
| 64 | n/a | if has_magic(basename): |
|---|
| 65 | n/a | if recursive and _isrecursive(basename): |
|---|
| 66 | n/a | glob_in_dir = _glob2 |
|---|
| 67 | n/a | else: |
|---|
| 68 | n/a | glob_in_dir = _glob1 |
|---|
| 69 | n/a | else: |
|---|
| 70 | n/a | glob_in_dir = _glob0 |
|---|
| 71 | n/a | for dirname in dirs: |
|---|
| 72 | n/a | for name in glob_in_dir(dirname, basename, dironly): |
|---|
| 73 | n/a | yield os.path.join(dirname, name) |
|---|
| 74 | n/a | |
|---|
| 75 | n/a | # These 2 helper functions non-recursively glob inside a literal directory. |
|---|
| 76 | n/a | # They return a list of basenames. _glob1 accepts a pattern while _glob0 |
|---|
| 77 | n/a | # takes a literal basename (so it only has to check for its existence). |
|---|
| 78 | n/a | |
|---|
| 79 | n/a | def _glob1(dirname, pattern, dironly): |
|---|
| 80 | n/a | names = list(_iterdir(dirname, dironly)) |
|---|
| 81 | n/a | if not _ishidden(pattern): |
|---|
| 82 | n/a | names = (x for x in names if not _ishidden(x)) |
|---|
| 83 | n/a | return fnmatch.filter(names, pattern) |
|---|
| 84 | n/a | |
|---|
| 85 | n/a | def _glob0(dirname, basename, dironly): |
|---|
| 86 | n/a | if not basename: |
|---|
| 87 | n/a | # `os.path.split()` returns an empty basename for paths ending with a |
|---|
| 88 | n/a | # directory separator. 'q*x/' should match only directories. |
|---|
| 89 | n/a | if os.path.isdir(dirname): |
|---|
| 90 | n/a | return [basename] |
|---|
| 91 | n/a | else: |
|---|
| 92 | n/a | if os.path.lexists(os.path.join(dirname, basename)): |
|---|
| 93 | n/a | return [basename] |
|---|
| 94 | n/a | return [] |
|---|
| 95 | n/a | |
|---|
| 96 | n/a | # Following functions are not public but can be used by third-party code. |
|---|
| 97 | n/a | |
|---|
| 98 | n/a | def glob0(dirname, pattern): |
|---|
| 99 | n/a | return _glob0(dirname, pattern, False) |
|---|
| 100 | n/a | |
|---|
| 101 | n/a | def glob1(dirname, pattern): |
|---|
| 102 | n/a | return _glob1(dirname, pattern, False) |
|---|
| 103 | n/a | |
|---|
| 104 | n/a | # This helper function recursively yields relative pathnames inside a literal |
|---|
| 105 | n/a | # directory. |
|---|
| 106 | n/a | |
|---|
| 107 | n/a | def _glob2(dirname, pattern, dironly): |
|---|
| 108 | n/a | assert _isrecursive(pattern) |
|---|
| 109 | n/a | yield pattern[:0] |
|---|
| 110 | n/a | yield from _rlistdir(dirname, dironly) |
|---|
| 111 | n/a | |
|---|
| 112 | n/a | # If dironly is false, yields all file names inside a directory. |
|---|
| 113 | n/a | # If dironly is true, yields only directory names. |
|---|
| 114 | n/a | def _iterdir(dirname, dironly): |
|---|
| 115 | n/a | if not dirname: |
|---|
| 116 | n/a | if isinstance(dirname, bytes): |
|---|
| 117 | n/a | dirname = bytes(os.curdir, 'ASCII') |
|---|
| 118 | n/a | else: |
|---|
| 119 | n/a | dirname = os.curdir |
|---|
| 120 | n/a | try: |
|---|
| 121 | n/a | with os.scandir(dirname) as it: |
|---|
| 122 | n/a | for entry in it: |
|---|
| 123 | n/a | try: |
|---|
| 124 | n/a | if not dironly or entry.is_dir(): |
|---|
| 125 | n/a | yield entry.name |
|---|
| 126 | n/a | except OSError: |
|---|
| 127 | n/a | pass |
|---|
| 128 | n/a | except OSError: |
|---|
| 129 | n/a | return |
|---|
| 130 | n/a | |
|---|
| 131 | n/a | # Recursively yields relative pathnames inside a literal directory. |
|---|
| 132 | n/a | def _rlistdir(dirname, dironly): |
|---|
| 133 | n/a | names = list(_iterdir(dirname, dironly)) |
|---|
| 134 | n/a | for x in names: |
|---|
| 135 | n/a | if not _ishidden(x): |
|---|
| 136 | n/a | yield x |
|---|
| 137 | n/a | path = os.path.join(dirname, x) if dirname else x |
|---|
| 138 | n/a | for y in _rlistdir(path, dironly): |
|---|
| 139 | n/a | yield os.path.join(x, y) |
|---|
| 140 | n/a | |
|---|
| 141 | n/a | |
|---|
| 142 | n/a | magic_check = re.compile('([*?[])') |
|---|
| 143 | n/a | magic_check_bytes = re.compile(b'([*?[])') |
|---|
| 144 | n/a | |
|---|
| 145 | n/a | def has_magic(s): |
|---|
| 146 | n/a | if isinstance(s, bytes): |
|---|
| 147 | n/a | match = magic_check_bytes.search(s) |
|---|
| 148 | n/a | else: |
|---|
| 149 | n/a | match = magic_check.search(s) |
|---|
| 150 | n/a | return match is not None |
|---|
| 151 | n/a | |
|---|
| 152 | n/a | def _ishidden(path): |
|---|
| 153 | n/a | return path[0] in ('.', b'.'[0]) |
|---|
| 154 | n/a | |
|---|
| 155 | n/a | def _isrecursive(pattern): |
|---|
| 156 | n/a | if isinstance(pattern, bytes): |
|---|
| 157 | n/a | return pattern == b'**' |
|---|
| 158 | n/a | else: |
|---|
| 159 | n/a | return pattern == '**' |
|---|
| 160 | n/a | |
|---|
| 161 | n/a | def escape(pathname): |
|---|
| 162 | n/a | """Escape all special characters. |
|---|
| 163 | n/a | """ |
|---|
| 164 | n/a | # Escaping is done by wrapping any of "*?[" between square brackets. |
|---|
| 165 | n/a | # Metacharacters do not work in the drive part and shouldn't be escaped. |
|---|
| 166 | n/a | drive, pathname = os.path.splitdrive(pathname) |
|---|
| 167 | n/a | if isinstance(pathname, bytes): |
|---|
| 168 | n/a | pathname = magic_check_bytes.sub(br'[\1]', pathname) |
|---|
| 169 | n/a | else: |
|---|
| 170 | n/a | pathname = magic_check.sub(r'[\1]', pathname) |
|---|
| 171 | n/a | return drive + pathname |
|---|