19
19
This module defines a class which creates proxy objects for regex
20
20
compilation. This allows overriding re.compile() to return lazily compiled
23
23
We do this rather than just providing a new interface so that it will also
24
24
be used by existing Python modules that create regexs.
42
42
class LazyRegex(object):
43
43
"""A proxy around a real regex, which won't be compiled until accessed."""
46
45
# These are the parameters on a real _sre.SRE_Pattern object, which we
47
46
# will map to local members so that we don't have the proxy overhead.
48
47
_regex_attributes_to_copy = [
49
'__copy__', '__deepcopy__', 'findall', 'finditer', 'match',
50
'scanner', 'search', 'split', 'sub', 'subn'
48
'__copy__', '__deepcopy__', 'findall', 'finditer', 'match',
49
'scanner', 'search', 'split', 'sub', 'subn'
53
52
# We use slots to keep the overhead low. But we need a slot entry for
54
53
# all of the attributes we will copy
55
54
__slots__ = ['_real_regex', '_regex_args', '_regex_kwargs',
56
] + _regex_attributes_to_copy
55
] + _regex_attributes_to_copy
58
def __init__(self, args=(), kwargs={}):
57
def __init__(self, args, kwargs):
59
58
"""Create a new proxy object, passing in the args to pass to re.compile
61
60
:param args: The `*args` to pass to re.compile
75
74
def _real_re_compile(self, *args, **kwargs):
76
75
"""Thunk over to the original re.compile"""
78
return _real_re_compile(*args, **kwargs)
77
return re.compile(*args, **kwargs)
79
78
except re.error as e:
80
79
# raise InvalidPattern instead of re.error as this gives a
81
80
# cleaner message to the user.
82
raise InvalidPattern('"' + args[0] + '" ' +str(e))
81
raise InvalidPattern('"' + args[0] + '" ' + str(e))
84
83
def __getstate__(self):
85
84
"""Return the state to use when pickling."""
112
111
:return: a LazyRegex proxy object.
114
113
return LazyRegex(args, kwargs)
117
def install_lazy_compile():
118
"""Make lazy_compile the default compile mode for regex compilation.
120
This overrides re.compile with lazy_compile. To restore the original
121
functionality, call reset_compile().
123
re.compile = lazy_compile
127
"""Restore the original function to re.compile().
129
It is safe to call reset_compile() multiple times, it will always
130
restore re.compile() to the value that existed at import time.
131
Though the first call will reset back to the original (it doesn't
134
re.compile = _real_re_compile
137
_real_re_compile = re.compile
138
if _real_re_compile is lazy_compile:
139
raise AssertionError(
140
"re.compile has already been overridden as lazy_compile, but this would" \
141
" cause infinite recursion")
144
# Some libraries calls re.finditer which fails it if receives a LazyRegex.
145
if getattr(re, 'finditer', False):
146
def finditer_public(pattern, string, flags=0):
147
if isinstance(pattern, LazyRegex):
148
return pattern.finditer(string)
150
return _real_re_compile(pattern, flags).finditer(string)
151
re.finditer = finditer_public
added
removed
breezy/lazy_regex.py
