TypeError when pickling directly instantiated urllib.parse result objects #160761

Python Version

Python 3.15 / main

Operating System

All platforms

Description

When directly instantiating result objects from urllib.parse (SplitResult, ParseResult, DefragResult, or their bytes counterparts), pickling or calling __getstate__() raises a TypeError:

import urllib.parse, pickle
res = urllib.parse.SplitResult('http', 'example.com', '/', '', '')
pickle.dumps(res)

Output:

TypeError: 'NoneType' object is not subscriptable

This occurs because _ResultBase.__getstate__() assumes super().__getstate__() returns a 2-tuple:

def __getstate__(self):
    state = super().__getstate__()
    try:
        if state[1]['_keep_empty'] == _MISSING_AS_NONE_DEFAULT:
            del state[1]['_keep_empty']
            if state == (None, {}):
                state = None
    except LookupError:
        pass
    return state

When _keep_empty has not been set (for instance, when constructing the result class directly rather than via urlsplit() or urlparse()), object.__getstate__() returns None. state[1] then fails with TypeError: 'NoneType' object is not subscriptable, which is not handled by except LookupError:.

Checking if state is not None: before subscripting resolves the error.

Linked PRs

• gh-160761: Fix TypeError when pickling directly instantiated urllib.parse result objects #160762