iri_to_uri undesirably converts ~ to %7E

[jille@…]

At the moment it is not possible to redirect to so-called user-dirs (e.g. ​http://host/~username/); because HttpResponseRedirect uses iri_to_uri, which does:

  return urllib.quote(smart_str(iri), safe='/#%[]=:;$&()+,!?*')

By adding ~ to the list of 'safe characters' my problem goes away.

My apologies if I ain't clear enough.

[jille@…]

Proposed patch

[James Bennett]

Per RFC 3986, /~username/ and /%7Eusername/ are equivalent, and two otherwise-equivalent URIs which differ only in whether they encode the tilde identify the same resource. Additionally, RFC 1738 declares the tilde "unsafe" and states that it must always be encoded within a URL. Thus, as far as I can tell, iri_to_uri is behaving correctly.

[jille@…]

Section 2.3 of RFC 3986 states:

  URIs that differ in the replacement of an unreserved character with
   its corresponding percent-encoded US-ASCII octet are equivalent: they
   identify the same resource. 

and

  For consistency, percent-encoded octets in the ranges of ALPHA
   (%41-%5A and %61-%7A), DIGIT (%30-%39), hyphen (%2D), period (%2E),
   underscore (%5F), or tilde (%7E) should not be created by URI
   producers

So, Django should not do it; however lighttpd (in my case) should accept it.
I'll file this as an lighttpd bug.

[jille@…]

FTR: Filed as ​http://redmine.lighttpd.net/issues/2124

[Luke Plant]

RFC 3986 obsoletes 1738, so we have to go with 3986 here, especially as it specifically addresses the issue of tilde, noting (in section 2.4) that older implementations might produce %7E, and says that URI producers should not be producing %7E (section 2.3), as jille@… noted above. So I'm re-opening.

[Gary Wilson]

django-dev discussion: ​http://groups.google.com/group/django-developers/t/281127e5c4aeef9d

[Gary Wilson]

(In [12066]) Fixed #12445 -- Added ' (single quote), @ (at sign), and ~ (tilde) to safe characters in iri_to_uri function.

[Gary Wilson]

(In [12067]) [1.1.X] Fixed #12445 -- Added ' (single quote), @ (at sign), and ~ (tilde) to safe characters in iri_to_uri function.

Backport of r12066 from trunk.