Inconsistent url parsing between urllib.parse and urllib.request with ":@" in netloc

Bug report

With "https://host:@1/path", urlib.parse.urlsplit parses "host" as the user and "1" as the hostname with no port.
If I use this url with urllib.request.urlopen, it parses "@1" as being the port.

Python 3.9.6 (default, Jun 28 2021, 08:57:49) 
[GCC 11.2.0] on linux
Type "help", "copyright", "credits" or "license" for more information.
>>> from urllib.parse import urlsplit
>>> s = urlsplit("https://host:@1/path")
>>> s.hostname
'1'
>>> s.port
>>> s.username
'host'
>>> import urllib.request
>>> urllib.request.urlopen(urllib.request.Request("https://host:@1/path"))
Traceback (most recent call last):
  File "/usr/lib/python3.9/http/client.py", line 872, in _get_hostport
ValueError: invalid literal for int() with base 10: '@1'

During handling of the above exception, another exception occurred:

Traceback (most recent call last):
  File "<stdin>", line 1, in <module>
  File "/usr/lib/python3.9/urllib/request.py", line 214, in urlopen
  File "/usr/lib/python3.9/urllib/request.py", line 517, in open
  File "/usr/lib/python3.9/urllib/request.py", line 534, in _open
  File "/usr/lib/python3.9/urllib/request.py", line 494, in _call_chain
  File "/usr/lib/python3.9/urllib/request.py", line 1389, in https_open
  File "/usr/lib/python3.9/urllib/request.py", line 1315, in do_open
  File "/usr/lib/python3.9/http/client.py", line 1384, in __init__
  File "/usr/lib/python3.9/http/client.py", line 834, in __init__
  File "/usr/lib/python3.9/http/client.py", line 877, in _get_hostport
http.client.InvalidURL: nonnumeric port: '@1'

Your environment

• CPython versions tested on: 3.9.6
• Operating system and architecture: Linux

Inizia con urllib.parse.urlsplit e urllib.request.urlopen, quindi traccia la gestione di host/port di http.client mostrata nel traceback. Confronta il modo in cui ciascun percorso interpreta https://host:@1/path e identifica il risultato coerente previsto; il lavoro è completato quando entrambi i percorsi concordano senza il parsing mismatch segnalato e i test pertinenti hanno esito positivo.