Allow image src without scheme and with external domain

[fabienheureux]

Description

If images have a src attribute like //example.com/image.jpg, the imported url does not care about the domain and prepend the wagtail domain, hence leading to a 404 when downloading the image

---

• Testing
  • CI passes
  • If necessary, tests are added for new or fixed behaviour
  • These changes do not reduce test coverage
• Documentation.
  • Documentation changes are not necessary because they do not change how the importer usage

[fabienheureux]

Sorry for the noise, I pushed without running tests locally, again 🙄, they now pass.

@nickmoreton do you think you could take a look ? I am not very proud of this approach, especially the hardcoded file extensions, but I did not see how could we better handle these various cases.

[nickmoreton]

@fabienheureux Thanks we really appreciate the pull request.

@nimasmi I tested it out and is working OK. Would you take a look please and confirm happy to merge. Thanks

[nimasmi]

Thanks @fabienheureux. I'm always in favour of urlparse over string manipulation. It looks like we should remove some further custom string parsing.

[nimasmi]

Issue: In the following case, the unrecognised local file extension is interpreted as a domain:

>>> get_absolute_src("image.bmp", "https://www.example.com")
'https://image.bmp'

urlparse results have a .hostname attribute that you can use in this situation. You will have to remove the .lstrip("/") line though.

>>> from urllib.parse import urlparse
>>> urlparse("image.bmp").hostname
>>> urlparse("example.com/image.bmp").hostname
>>> urlparse("//example.com/image.bmp").hostname
'example.com'

[nimasmi]

Suggestion: Add a test case for an unrecognised extension on the local domain: get_absolute_src("/fakeimage.badextension", "http://www.example.com")

[fabienheureux]

@nickmoreton I am cleaning up old PRs.

Can I close this one ?

I hope I could have take @nimasmi comment into account but never found the time and I won't use this plugin in a near future.