| 1 |
import os |
| 2 |
import re |
| 3 |
|
| 4 |
_default_seps = os.sep + str(os.altsep) * bool(os.altsep) |
| 5 |
|
| 6 |
|
| 7 |
class Translator: |
| 8 |
""" |
| 9 |
>>> Translator('xyz') |
| 10 |
Traceback (most recent call last): |
| 11 |
... |
| 12 |
AssertionError: Invalid separators |
| 13 |
|
| 14 |
>>> Translator('') |
| 15 |
Traceback (most recent call last): |
| 16 |
... |
| 17 |
AssertionError: Invalid separators |
| 18 |
""" |
| 19 |
|
| 20 |
seps: str |
| 21 |
|
| 22 |
def __init__(self, seps: str = _default_seps): |
| 23 |
assert seps and set(seps) <= set(_default_seps), "Invalid separators" |
| 24 |
self.seps = seps |
| 25 |
|
| 26 |
def translate(self, pattern): |
| 27 |
""" |
| 28 |
Given a glob pattern, produce a regex that matches it. |
| 29 |
""" |
| 30 |
return self.extend(self.match_dirs(self.translate_core(pattern))) |
| 31 |
|
| 32 |
def extend(self, pattern): |
| 33 |
r""" |
| 34 |
Extend regex for pattern-wide concerns. |
| 35 |
|
| 36 |
Apply '(?s:)' to create a non-matching group that |
| 37 |
matches newlines (valid on Unix). |
| 38 |
|
| 39 |
Append '\z' to imply fullmatch even when match is used. |
| 40 |
""" |
| 41 |
return rf'(?s:{pattern})\z' |
| 42 |
|
| 43 |
def match_dirs(self, pattern): |
| 44 |
""" |
| 45 |
Ensure that zipfile.Path directory names are matched. |
| 46 |
|
| 47 |
zipfile.Path directory names always end in a slash. |
| 48 |
""" |
| 49 |
return rf'{pattern}[/]?' |
| 50 |
|
| 51 |
def translate_core(self, pattern): |
| 52 |
r""" |
| 53 |
Given a glob pattern, produce a regex that matches it. |
| 54 |
|
| 55 |
>>> t = Translator() |
| 56 |
>>> t.translate_core('*.txt').replace('\\\\', '') |
| 57 |
'[^/]*\\.txt' |
| 58 |
>>> t.translate_core('a?txt') |
| 59 |
'a[^/]txt' |
| 60 |
>>> t.translate_core('**/*').replace('\\\\', '') |
| 61 |
'.*/[^/][^/]*' |
| 62 |
""" |
| 63 |
self.restrict_rglob(pattern) |
| 64 |
return ''.join(map(self.replace, separate(self.star_not_empty(pattern)))) |
| 65 |
|
| 66 |
def replace(self, match): |
| 67 |
""" |
| 68 |
Perform the replacements for a match from :func:`separate`. |
| 69 |
""" |
| 70 |
return match.group('set') or ( |
| 71 |
re.escape(match.group(0)) |
| 72 |
.replace('\\*\\*', r'.*') |
| 73 |
.replace('\\*', rf'[^{re.escape(self.seps)}]*') |
| 74 |
.replace('\\?', r'[^/]') |
| 75 |
) |
| 76 |
|
| 77 |
def restrict_rglob(self, pattern): |
| 78 |
""" |
| 79 |
Raise ValueError if ** appears in anything but a full path segment. |
| 80 |
|
| 81 |
>>> Translator().translate('**foo') |
| 82 |
Traceback (most recent call last): |
| 83 |
... |
| 84 |
ValueError: ** must appear alone in a path segment |
| 85 |
""" |
| 86 |
seps_pattern = rf'[{re.escape(self.seps)}]+' |
| 87 |
segments = re.split(seps_pattern, pattern) |
| 88 |
if any('**' in segment and segment != '**' for segment in segments): |
| 89 |
raise ValueError("** must appear alone in a path segment") |
| 90 |
|
| 91 |
def star_not_empty(self, pattern): |
| 92 |
""" |
| 93 |
Ensure that * will not match an empty segment. |
| 94 |
""" |
| 95 |
|
| 96 |
def handle_segment(match): |
| 97 |
segment = match.group(0) |
| 98 |
return '?*' if segment == '*' else segment |
| 99 |
|
| 100 |
not_seps_pattern = rf'[^{re.escape(self.seps)}]+' |
| 101 |
return re.sub(not_seps_pattern, handle_segment, pattern) |
| 102 |
|
| 103 |
|
| 104 |
def separate(pattern): |
| 105 |
""" |
| 106 |
Separate out character sets to avoid translating their contents. |
| 107 |
|
| 108 |
>>> [m.group(0) for m in separate('*.txt')] |
| 109 |
['*.txt'] |
| 110 |
>>> [m.group(0) for m in separate('a[?]txt')] |
| 111 |
['a', '[?]', 'txt'] |
| 112 |
""" |
| 113 |
return re.finditer(r'([^\[]+)|(?P<set>[\[].*?[\]])|([\[][^\]]*$)', pattern) |