Strip leading . from cookie domain names.

[icemac]

If the value of the domain attribute in the cookie starts with . Zope currently breaks.
But according to https://www.rfc-editor.org/rfc/rfc6265#section-4.1.2.3 it should be allowed.

I added a test and a fix, without the fix the new test breaks with:

Error in test test_domain (ZPublisher.tests.test_cookie.CookieParameterRegistryTests)
Traceback (most recent call last):
  File "/.../lib/python3.7/unittest/case.py", line 59, in testPartExecutor
    yield
  File "/.../lib/python3.7/unittest/case.py", line 628, in run
    testMethod()
  File "/.../Zope/src/ZPublisher/tests/test_cookie.py", line 150, in test_domain
    _, v = convertCookieParameter("domain", ".zope.dev")
  File "/.../Zope/src/ZPublisher/cookie.py", line 152, in convert
    return nn, (value if value is None else self[nn](value))
  File "/.../Zope/src/ZPublisher/cookie.py", line 249, in domain_converter
    value = ".".join(to_str(ToASCII(nameprep(c))) for c in u_value.split("."))
  File "/.../Zope/src/ZPublisher/cookie.py", line 249, in <genexpr>
    value = ".".join(to_str(ToASCII(nameprep(c))) for c in u_value.split("."))
  File "/.../lib/python3.7/encodings/idna.py", line 73, in ToASCII
    raise UnicodeError("label empty or too long")
UnicodeError: label empty or too long

[d-maurer]

Michael Howitz wrote at 2022-5-17 23:25 -0700: ***@***.*** commented on this pull request.

> @@ -242,6 +242,10 @@ def domain_converter(value): return value from encodings.idna import ToASCII from encodings.idna import nameprep

***@***.*** Is it necessary to do these imports here instead of at the top of the module? They are here for documentation/modularity purposes: this documents that they are special to this converter and not used elsewhere. Maybe, we should consider "#1033": meanwhile, I believe that using the IDNA encoding (in this place) is wrong altogether and that it should get replaced by normal URL quoting (with safe characters defined by "https://datatracker.ietf.org/doc/html/rfc3986.html#section-3.2.2"). IDNA is likely necessary only in interactions with the DNS; Browsers apparently do not expect IDNA encoding in the `host` part of an URL but normal URL quoting.

[dataflake]

I'm going to merge this, #1033 is a separate issue for a separate PR.