views:

204

answers:

2
+3  Q: 

Django DRY Feeds

I'm using the Django Feeds Framework and it's really nice, very intuitive and easy to use. But, I think there is a problem when creating links to feeds in HTML.

For example:

<link rel="alternate" type="application/rss+xml" title="{{ feed_title }}" href="{{ url_of_feed }}" />

Link's HREF attribute can be easily found out, just use reverse()

But, what about the TITLE attribute? Where the template engine should look for this? Even more, what if the feed is build up dinamically and the title depends on parameters (like this)?

I can't come up with a solution that "seems" DRY to me... All that I can come up with is using context processors o template tags, but it gets messy when the context procesor/template tag has to find parameters to construct the Feed class, and writing this I realize I don't even know how to create a Feed instance myself within the view.

If I put all this logic in the view, it would not be just one view. Also, the value for TITLE would be in the view AND in the feed.

A: 

Just a guess (as I have not used feeds yet in my django app), but you could add a special template_context for your feed with your feed object and use it in your base.html.

xyld
I have think on this, but this aproach still has problems, like how to obtain the current category. Context processors can't access context data, all they take is the Request object.
Mandx
A: 

I'm not fully satisfied with this solution, it may break feeds using Request and depends on a magic method. There it goes:

#coding:utf-8
# Author: Armando Pérez Marqués <[email protected]>
# Purpose: Django TemplateTag to output feed links in templates in a DRY way
# Created: 05/07/2010

import re

from django import template
from django.conf import settings
from django.contrib.syndication.views import Feed
from django.core.urlresolvers import reverse, resolve, NoReverseMatch
from django.template import Node
from django.template import TemplateSyntaxError
from django.utils.encoding import smart_str
from django.utils.html import escape as html_escape
from django.utils.safestring import mark_safe

register = template.Library()

kwarg_re = re.compile(r"(?:(\w+)=)?(.+)")

class FeedInfoNode(Node):
    def __init__(self, view_name, args, kwargs, asvar):
        self.view_name = view_name
        self.args = args
        self.kwargs = kwargs
        self.asvar = asvar

    def render(self, context):
        args = [arg.resolve(context) for arg in self.args]
        kwargs = dict([(smart_str(k,'ascii'), v.resolve(context))
                       for k, v in self.kwargs.items()])

        # Try to look up the URL twice: once given the view name, and again
        # relative to what we guess is the "main" app. If they both fail,
        # re-raise the NoReverseMatch unless we're using the
        # {% feed_info ... as var %} construct in which cause return nothing.
        url = ''
        try:
            url = reverse(self.view_name, args=args, kwargs=kwargs, current_app=context.current_app)
        except NoReverseMatch, e:
            if settings.SETTINGS_MODULE:
                project_name = settings.SETTINGS_MODULE.split('.')[0]
                try:
                    url = reverse(project_name + '.' + self.view_name,
                              args=args, kwargs=kwargs, current_app=context.current_app)
                except NoReverseMatch:
                    if self.asvar is None:
                        # Re-raise the original exception, not the one with
                        # the path relative to the project. This makes a
                        # better error message.
                        raise e
            else:
                if self.asvar is None:
                    raise e

        if 'request' in context:
            request = context['request']
        else:
            request = None

        feed_instance, feed_args, feed_kwargs = resolve(url)
        if not isinstance(feed_instance, Feed):
            raise NoReverseMatch, \
                  'feed_info can only reverse class-based feeds'

        feed_obj = feed_instance.get_object(request, *feed_args, **feed_kwargs)

        feed_data = {
            'url': url,
            'obj': feed_instance,
            'args': feed_args,
            'kwargs': feed_kwargs,
            #'title': html_escape(feed_instance.__get_dynamic_attr('title', obj)),
            'title': html_escape(
                feed_instance._Feed__get_dynamic_attr('title', feed_obj)
                ),
            'type': feed_instance.feed_type.mime_type,
        }

        if self.asvar:
            context[self.asvar] = feed_data
            return ''
        else:
            return mark_safe(
                '<link rel="alternate" type="%(type)s" title="%(title)s" href="%(url)s" />' \
                % feed_data
            )

def feed_info(parser, token):
    """
    Returns an mapping containing populated info about the reversed feed
    Works exactly as the url tag, but the mapping is not returned, instead
    a variable is always set  in the context.

    This is a way to define links that aren't tied to a particular URL
    configuration::

        {% feed_info path.to.some_feed_view_class arg1 arg2 as feed_info_var %}

        or

        {% feed_info path.to.some_feed_view_class name1=value1 name2=value2 as feed_info_var %}
    """

    bits = token.split_contents()
    if len(bits) < 2:
        raise TemplateSyntaxError("'%s' takes at least one argument"
                                  " (path to a feed view)" % bits[0])
    viewname = bits[1]
    args = []
    kwargs = {}
    asvar = None
    bits = bits[2:]
    if len(bits) >= 2 and bits[-2] == 'as':
        asvar = bits[-1]
        bits = bits[:-2]

    # Backwards compatibility: check for the old comma separated format
    # {% url urlname arg1,arg2 %}
    # Initial check - that the first space separated bit has a comma in it
    if bits and ',' in bits[0]:
        check_old_format = True
        # In order to *really* be old format, there must be a comma
        # in *every* space separated bit, except the last.
        for bit in bits[1:-1]:
            if ',' not in bit:
                # No comma in this bit. Either the comma we found
                # in bit 1 was a false positive (e.g., comma in a string),
                # or there is a syntax problem with missing commas
                check_old_format = False
                break
    else:
        # No comma found - must be new format.
        check_old_format = False

    if check_old_format:
        # Confirm that this is old format by trying to parse the first
        # argument. An exception will be raised if the comma is
        # unexpected (i.e. outside of a static string).
        match = kwarg_re.match(bits[0])
        if match:
            value = match.groups()[1]
            try:
                parser.compile_filter(value)
            except TemplateSyntaxError:
                bits = ''.join(bits).split(',')

    # Now all the bits are parsed into new format,
    # process them as template vars
    if len(bits):
        for bit in bits:
            match = kwarg_re.match(bit)
            if not match:
                raise TemplateSyntaxError("Malformed arguments to url tag")
            name, value = match.groups()
            if name:
                kwargs[name] = parser.compile_filter(value)
            else:
                args.append(parser.compile_filter(value))

    return FeedInfoNode(viewname, args, kwargs, asvar)

feed_info = register.tag(feed_info)

I'm starting with the code of the {% url %} template tag, and then, after obtaining the feed's URL, use resolve() to get the Feed subclass instance, and then get the needed attributes.

Caveats

  • Requires Django 1.2 Class Feeds, don't know exactly how to do this with the old way of feeds.
  • If the feed class uses the request object, the request context processor must be configured, since None is passed if it isn't present in the context.
  • There's an oddity with Feed.__get_dynamic_attr(). The Feed subclass instance doesn't have this method; instead, it appears with another name. Don't know how to figure the name out at runtime...
Mandx
I'm going with this, since it's analogous to use `{% url %}`
Mandx