[fix] Fix domain detection

[Bixilon]

Should fix #273, not tested

[nemesifier]

Can you explain this change with a few examples of why the current way is not working and the new way is better, please?

[Bixilon]

I have no clue what the library is doing, but it should cut off the first part of a domain (test.abc.de -> abc.de). The library just does it with the top level domain (test.abc.de -> de, but not even then correct, check a domain with a dash test.ab-cd.de, it then just returns nothing).

It is used in a header (nginx), and if that header is not set correctly, things like chart loading fail.

[Bixilon]

Tested it, working fine!

[Bixilon]

Any chance of getting this merged?

[nemesifier]

Testing with tldextract-5.1.2:

In [9]: print(tldextract.extract('openwisp.org').registered_domain)
openwisp.org

In [10]: print(tldextract.extract('api.openwisp.org').registered_domain)
openwisp.org

In [11]: print(tldextract.extract('test.api.openwisp.org').registered_domain)
openwisp.org

In [12]: print(tldextract.extract('test.abc.de').registered_domain)
abc.de

In [13]: print(tldextract.extract('abc.de').registered_domain)
abc.de

In [14]: print(tldextract.extract('test.test.abc.de').registered_domain)
abc.de

I cannot replicate the issue you are mentioning here. I will try to update this library and modify the different parts of the code to make sure they're consistent.

[Bixilon]

try extracting from api.open-wisp.de

[nemesifier]

try extracting from api.open-wisp.de

As far as I know, dashes are not allowed in the domain.
Yes they are allowed.

try extracting from api.open-wisp.de

In [1]: import tldextract

In [2]: ext=tldextract.extract('api.open-wisp.org')

In [3]: ext
Out[3]: ExtractResult(subdomain='api', domain='open-wisp', suffix='org', is_private=False)

In [4]: ext.registered_domain
Out[4]: 'open-wisp.org'

What's the issue? Please update your local copy of the library.
PR in #300.