Another way to extend (specifically meaning, add new methods, not change existing ones) classes, even built-in ones, is to use a preprocessor that adds the ability to extend out of/above the scope of Python itself, converting the extension to normal Python syntax before Python actually gets to see it.
I've done this to extend Python 2's str() class, for instance. str() is a particularly interesting target because of the implicit linkage to quoted data such as 'this' and 'that'.
Here's some extending code, where the only added non-Python syntax is the extend:testDottedQuad bit:
extend:testDottedQuad
def testDottedQuad(strObject):
if not isinstance(strObject, basestring): return False
listStrings = strObject.split('.')
if len(listStrings) != 4: return False
for strNum in listStrings:
try: val = int(strNum)
except: return False
if val < 0: return False
if val > 255: return False
return True
After which I can write in the code fed to the preprocessor:
if '192.168.1.100'.testDottedQuad():
doSomething()
dq = '216.126.621.5'
if not dq.testDottedQuad():
throwWarning();
dqt = ''.join(['127','.','0','.','0','.','1']).testDottedQuad()
if dqt:
print 'well, that was fun'
The preprocessor eats that, spits out normal Python without monkeypatching, and Python does what I intended it to do.
Just as a c preprocessor adds functionality to c, so too can a Python preprocessor add functionality to Python.
My preprocessor implementation is too large for a stack overflow answer, but for those who might be interested, it is here on GitHub.