blob: f8f0701f9cf451f881397ae0ef4bb8185530dbe3 (
plain)
1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
24
25
26
27
28
29
30
31
32
33
34
35
36
|
from django.contrib.syndication.feeds import * # merge modules
import sys
from django.contrib.syndication.feeds import Feed as DjangoFeed
from coffin.template import loader as coffin_loader
class Feed(DjangoFeed):
"""A ``Feed`` implementation that renders it's title and
description templates using Jinja2.
Unfortunately, Django's base ``Feed`` class is not very extensible
in this respect at all. For a real solution, we'd have to essentially
have to duplicate the whole class. So for now, we use this terrible
non-thread safe hack.
Another, somewhat crazy option would be:
* Render the templates ourselves through Jinja2 (possible
introduce new attributes to avoid having to rewrite the
existing ones).
* Make the rendered result available to Django/the superclass by
using a custom template loader using a prefix, say
"feed:<myproject.app.views.MyFeed>". The loader would simply
return the Jinja-rendered template (escaped), the Django template
mechanism would find no nodes and just pass the output through.
Possible even worse than this though.
"""
def get_feed(self, *args, **kwargs):
parent_module = sys.modules[DjangoFeed.__module__]
old_loader = parent_module.loader
parent_module.loader = coffin_loader
try:
return super(Feed, self).get_feed(*args, **kwargs)
finally:
parent_module.loader = old_loader
|
