Repository navigation
Can't handle URLs with @ character in userinfo section #328
Description
>>> import httpx
>>> httpx.URL("https://user@gmail.com:password@google.com")
Traceback (most recent call last):
File "<stdin>", line 1, in <module>
File "/home/sethmlarson/Desktop/http3/httpx/models.py", line 112, in __init__
raise InvalidURL("No host included in URL.")
httpx.exceptions.InvalidURL: No host included in URL.This is a common use-case for proxies and websites using email addresses as usernames for authentication.
This might require a change in rfc3986 or the way we parse URLs. :)
Reacted by Kim Christie
Activity
cansarigol commentedon Sep 9, 2019
Hi @sethmlarson I've pushed a PR about this topic. Could your review it?
lovelydinosaur commentedon Jul 27, 2020
So, something I think we may want to re-review prior to 1.0 is which internal URL parsing implementation is really our best option. This issue (and perhaps also #858, although maybe that's exclusively an issue on our side?) currently means that we're a little less robust on URL parsing than the wonderfully battle hardened
urllib3, which is something we ought to resolve.Making a change here wouldn't have any affect on our public API. That's tied down and looks that way we want. But this could potentially change which dependancies we choose to ship with.
Some options here are...
rfc3986. Because we're doing just fine really. Let's just make sure we're contributing our time to helping getting this issue resolved.irl.urllib3implementation, possibly by vendoring just the URL parsing into a (properly licensed and credited) slimline stand-alone package that just implements plain functions forurl_parseandurl_join, that only deal with super plain named-tuple-like arguments.Of those three options, the last seems like it'd be the most robust approach, since
urllib3have been doing this for years. Having said that, I don't really have enough context about which of these three options theurllib3team themselves would consider to be most suitable.No doubt @sigmavirus24 and @sethmlarson both have far more insight into this, tho I don't want to push for anyone's time here.