17
17
"""Lazily compiled regex objects.
19
This module defines a class which creates proxy objects for regex
20
compilation. This allows overriding re.compile() to return lazily compiled
23
We do this rather than just providing a new interface so that it will also
24
be used by existing Python modules that create regexs.
19
This module defines a class which creates proxy objects for regex compilation.
20
This allows overriding re.compile() to return lazily compiled objects.
27
from __future__ import absolute_import
34
class InvalidPattern(errors.BzrError):
36
_fmt = ('Invalid pattern(s) found. %(msg)s')
38
def __init__(self, msg):
25
from bzrlib import errors
42
28
class LazyRegex(object):
43
29
"""A proxy around a real regex, which won't be compiled until accessed."""
45
32
# These are the parameters on a real _sre.SRE_Pattern object, which we
46
33
# will map to local members so that we don't have the proxy overhead.
47
34
_regex_attributes_to_copy = [
48
'__copy__', '__deepcopy__', 'findall', 'finditer', 'match',
49
'scanner', 'search', 'split', 'sub', 'subn'
35
'__copy__', '__deepcopy__', 'findall', 'finditer', 'match',
36
'scanner', 'search', 'split', 'sub', 'subn'
52
39
# We use slots to keep the overhead low. But we need a slot entry for
53
40
# all of the attributes we will copy
54
41
__slots__ = ['_real_regex', '_regex_args', '_regex_kwargs',
55
] + _regex_attributes_to_copy
42
] + _regex_attributes_to_copy
57
def __init__(self, args, kwargs):
44
def __init__(self, args=(), kwargs={}):
58
45
"""Create a new proxy object, passing in the args to pass to re.compile
60
:param args: The `*args` to pass to re.compile
61
:param kwargs: The `**kwargs` to pass to re.compile
47
:param args: The *args to pass to re.compile
48
:param kwargs: The **kwargs to pass to re.compile
63
50
self._real_regex = None
64
51
self._regex_args = args
74
61
def _real_re_compile(self, *args, **kwargs):
75
62
"""Thunk over to the original re.compile"""
77
return re.compile(*args, **kwargs)
64
return _real_re_compile(*args, **kwargs)
79
66
# raise InvalidPattern instead of re.error as this gives a
80
67
# cleaner message to the user.
81
raise InvalidPattern('"' + args[0] + '" ' + str(e))
83
def __getstate__(self):
84
"""Return the state to use when pickling."""
86
"args": self._regex_args,
87
"kwargs": self._regex_kwargs,
90
def __setstate__(self, dict):
91
"""Restore from a pickled state."""
92
self._real_regex = None
93
setattr(self, "_regex_args", dict["args"])
94
setattr(self, "_regex_kwargs", dict["kwargs"])
68
raise errors.InvalidPattern('"' + args[0] + '" ' +str(e))
96
70
def __getattr__(self, attr):
97
71
"""Return a member from the proxied regex object.
111
85
:return: a LazyRegex proxy object.
113
87
return LazyRegex(args, kwargs)
90
def install_lazy_compile():
91
"""Make lazy_compile the default compile mode for regex compilation.
93
This overrides re.compile with lazy_compile. To restore the original
94
functionality, call reset_compile().
96
re.compile = lazy_compile
100
"""Restore the original function to re.compile().
102
It is safe to call reset_compile() multiple times, it will always
103
restore re.compile() to the value that existed at import time.
104
Though the first call will reset back to the original (it doesn't
107
re.compile = _real_re_compile
110
_real_re_compile = re.compile
111
if _real_re_compile is lazy_compile:
112
raise AssertionError(
113
"re.compile has already been overridden as lazy_compile, but this would" \
114
" cause infinite recursion")