|
 |
9be401 |
#! /usr/bin/env python3
|
|
|
9be401 |
|
|
|
9be401 |
from __future__ import print_function
|
|
|
9be401 |
|
|
|
9be401 |
import re
|
|
|
9be401 |
|
|
|
9be401 |
try:
|
|
|
9be401 |
basestring
|
|
|
9be401 |
except NameError:
|
|
|
9be401 |
basestring = str
|
|
|
9be401 |
|
|
|
9be401 |
# ---------------------------------------------------------- Regex Match()
|
|
|
9be401 |
# beware, stupid python interprets backslashes in replace-parts only partially!
|
|
|
9be401 |
class MatchReplace:
|
|
|
9be401 |
""" A MatchReplace is a mix of a Python Pattern and a Replace-Template """
|
|
|
9be401 |
def __init__(self, matching, template, count = 0, flags = None):
|
|
|
9be401 |
""" setup a substition from regex 'matching' into 'template',
|
|
|
9be401 |
the replacement count default of 0 will replace all occurrences.
|
|
|
9be401 |
The first argument may be a Match object or it is a string that
|
|
|
9be401 |
will be turned into one by using Match(matching, flags). """
|
|
|
9be401 |
self.template = template
|
|
|
9be401 |
MatchReplace.__call__(self, matching, template, count, flags)
|
|
|
9be401 |
def __call__(self, matching, template = None, count = 0, flags = None):
|
|
|
9be401 |
""" other than __init__ the template may be left off to be unchanged"""
|
|
|
9be401 |
if isinstance(count, basestring): # count/flags swapped over?
|
|
|
9be401 |
flags = count; count = 0
|
|
|
9be401 |
if isinstance(matching, Match):
|
|
|
9be401 |
self.matching = matching
|
|
|
9be401 |
else:
|
|
|
9be401 |
self.matching = Match()(matching, flags) ## python 2.4.2 bug
|
|
|
9be401 |
if template is not None:
|
|
|
9be401 |
self.template = template
|
|
|
9be401 |
self.count = count
|
|
|
9be401 |
def __and__(self, string):
|
|
|
9be401 |
""" z = MatchReplace('foo', 'bar') & 'foo'; assert z = 'bar' """
|
|
|
9be401 |
text, self.matching.replaced = \
|
|
|
9be401 |
self.matching.regex.subn(self.template, string, self.count)
|
|
|
9be401 |
return text
|
|
|
9be401 |
def __rand__(self, string):
|
|
|
9be401 |
""" z = 'foo' & Match('foo') >> 'bar'; assert z = 'bar' """
|
|
|
9be401 |
text, self.matching.replaced = \
|
|
|
9be401 |
self.matching.regex.subn(self.template, string, self.count)
|
|
|
9be401 |
return text
|
|
|
9be401 |
def __iand__(self, string):
|
|
|
9be401 |
""" x = 'foo' ; x &= Match('foo') >> 'bar'; assert x == 'bar' """
|
|
|
9be401 |
string, self.matching.replaced = \
|
|
|
9be401 |
self.matching.regex.subn(self.template, string, self.count)
|
|
|
9be401 |
return string
|
|
|
9be401 |
def __rshift__(self, count):
|
|
|
9be401 |
" shorthand to set the replacement count: Match('foo') >> 'bar' >> 1 "
|
|
|
9be401 |
self.count = count ; return self
|
|
|
9be401 |
def __rlshift__(self, count):
|
|
|
9be401 |
self.count = count ; return self
|
|
|
9be401 |
|
|
|
9be401 |
class Match:
|
|
|
9be401 |
""" A Match is actually a mix of a Python Pattern and MatchObject """
|
|
|
9be401 |
def __init__(self, pattern = None, flags = None):
|
|
|
9be401 |
""" flags is a string: 'i' for case-insensitive etc.; it is just
|
|
|
9be401 |
short for a regex prefix: Match('foo','i') == Match('(?i)foo') """
|
|
|
9be401 |
Match.__call__(self, pattern, flags)
|
|
|
9be401 |
def __call__(self, pattern, flags = None):
|
|
|
9be401 |
assert isinstance(pattern, str) or pattern is None
|
|
|
9be401 |
assert isinstance(flags, str) or flags is None
|
|
|
9be401 |
self.replaced = 0 # set by subn() inside MatchReplace
|
|
|
9be401 |
self.found = None # set by search() to a MatchObject
|
|
|
9be401 |
self.pattern = pattern
|
|
|
9be401 |
if pattern is not None:
|
|
|
9be401 |
if flags:
|
|
|
9be401 |
self.regex = re.compile("(?"+flags+")"+self.pattern)
|
|
|
9be401 |
else:
|
|
|
9be401 |
self.regex = re.compile(self.pattern)
|
|
|
9be401 |
return self
|
|
|
9be401 |
def __repr__(self):
|
|
|
9be401 |
return self.pattern
|
|
|
9be401 |
def __truth__(self):
|
|
|
9be401 |
return self.found is not None
|
|
|
9be401 |
def __and__(self, string):
|
|
|
9be401 |
self.found = self.regex.search(string)
|
|
|
9be401 |
return self.__truth__()
|
|
|
9be401 |
def __rand__(self, string):
|
|
|
9be401 |
self.found = self.regex.search(string)
|
|
|
9be401 |
return self.__truth__()
|
|
|
9be401 |
def __rshift__(self, template):
|
|
|
9be401 |
return MatchReplace(self, template)
|
|
|
9be401 |
def __rlshift__(self, template):
|
|
|
9be401 |
return MatchReplace(self, template)
|
|
|
9be401 |
def __getitem__(self, index):
|
|
|
9be401 |
return self.group(index)
|
|
|
9be401 |
def group(self, index):
|
|
|
9be401 |
assert self.found is not None
|
|
|
9be401 |
return self.found.group(index)
|
|
|
9be401 |
def finditer(self, string):
|
|
|
9be401 |
return self.regex.finditer(string)
|
|
|
9be401 |
|
|
|
9be401 |
if __name__ == "__main__":
|
|
|
9be401 |
# matching:
|
|
|
9be401 |
if "foo" & Match("oo"):
|
|
|
9be401 |
print("oo")
|
|
|
9be401 |
x = Match()
|
|
|
9be401 |
if "foo" & x("(o+)"):
|
|
|
9be401 |
print(x[1])
|
|
|
9be401 |
# replacing:
|
|
|
9be401 |
y = "fooboo" & Match("oo") >> "ee"
|
|
|
9be401 |
print(y)
|
|
|
9be401 |
r = Match("oo") >> "ee"
|
|
|
9be401 |
print("fooboo" & r)
|
|
|
9be401 |
s = MatchReplace("oo", "ee")
|
|
|
9be401 |
print("fooboo" & s)
|