Filter entries by language

[memee]

While adopting feincms.translations.TranslatedObjectMixin, feincms.translations.Translation to my models it's easy for me to list them and pick proper translation with object.translation and then for example object.translation.title. I always had proper translation.

How to do the same with Elephantblog? I've registered 'navigation' translations extension: I can edit translations. But when i get object_list in template it's collection of all entries disregarding language.

[sbaechler]

You need to register the feincms.module.extensions.translations extension to make it work.

http://feincms-elephantblog.readthedocs.org/en/latest/installation.html

[saboter]

Hi, i have similar problem. I integrated elephantblog with use of application content in existing feincms project. The translations extension is properly registered but i still get all entries in object_list in entry archive regardless on current language code.

[simonkern]

Hey, I think i have a solution for your problem ;-) I am using kind of a language switch at the root level and create seperate page trees for every language. This solution works fine for me ;-)

just edit entry_archive.html template and put sth like the following in it:

"""

{% block object_list %}
{% for entry in object_list %}
#Only use those entries that are written in the current language
{% get_current_language as LANGUAGE_CODE %} 
{% if entry.language == LANGUAGE_CODE %}           
   

        
{{ entry }}
        
            {% for category in entry.fetched_categories %}
                {{ category }}{% if not forloop.last %},{% else %} | {% endif %}
            {% endfor %}
            {{ entry.published_on|date:"j F Y" }}
        
        {{ entry.first_image.render }}
        {{ entry.first_richtext.render }}
    
{% endif %}   
{% endfor %} 
{% endblock %}

"""

Sorry I don't know how to make github show the html tags as code :(

Best, Simon

[saboter]

Hello,

thanks for your advice. I fixed the problem in my fork with adding filtering by language:

https://github.com/saboter/feincms-elephantblog/commit/776a57f26d265fafe64c14d83eb364c3a9506cd6

[matthiask]

@saboter this solution isn't always correct. I'd recommend against adding this to the core. Most of the time the entries should be filtered by language (as you write) but I'm not sure that this is always the case.

If you need this, it's easy to add a filter to active_filters:

from elephantblog.models import Entry
from feincms.translations import short_language_code

Entry.objects.add_to_active_filters(
    lambda queryset: queryset.filter(language__startswith=short_language_code()),
    key='active_language')

Something like that should definitely be added to the documentation.

[saboter]

@matthiask very elegant and works like charm. Thank you. Should be definitely mentioned in docs.

[sbaechler]

It would actually make sense if added to the translation extension.

[matthiask]

@sbaechler no -- pages from feincms.module.page in other languages should not be considered inactive under all circumstances.

[sbaechler]

How about this:

https://github.com/sbaechler/feincms-elephantblog/commit/d49e4191ca82f37411d3d6be22df0b4dcf4f7920

[matthiask]

I have been thinking long and hard about this issue in the last few days, and have also discussed this offline with @sbaechler

• It's not correct to mark entries in other languages as inactive by adding the code snippet I posted. Just as an example, you'd want all blog entries to appear in /sitemap.xml, not just those of the active language.
• It's very fundamentally a decision of the view code. However, fumbling around in ElephantBlog.get_queryset from the outside is really ugly.

I've added some code in my fork because I really need this feature today. I'm not happy with the solution though, I'd like to avoid stuffing knowledge about extensions inside the view code if possible.

For the moment I'd recommend using the snippet above and mark it with a big warning NOT REALLY ACTIVE/INACTIVE.

[sbaechler]

I think putting the language filtering in the view code is the right approach. I'm concerned about thread-safety if it is put in the manager. The problem now is finding out if the translation extension is used or not.

About your proposal: Is the queryset not executed on the transform call? Then you could save a query if you check for the language attribute on the Entry model and then define the right filter method afterwards.

I also think TRANSLATION_STRATEGY should be an attribute of the ElephantblogMixin. This way you can override it in your view.

[matthiask]

I just had a different idea: What if we add a method which returns a set of URL patterns for some given configuration? That way you could do something like this:

from elephantblog.urls import elephantblog_patterns

urlpatterns = elephantblog_patterns(mixin=CustomMixin, list_view_kwargs={'only_active_language': True})

(Start is here: https://github.com/matthiask/feincms-elephantblog/commit/6756b5172f5270b56c35b71f744d1fd9a38fab05)

What do you think?

[sbaechler]

This is better but it would mean that you have to add Translations to two places. The register extensions and the urls. More elegant would be if it was enough to just register the translation extension and that would be it.

Can we put the language filtering into an ActiveLanguageMixin? To separate it from the main view code and easier reuse.

[simonkern]

Hey everyone, I just played around a bit and got the following solution: I am pretty new to Django and Python so forgive me if it's really stupid :-p

I just added another EntryManager to models.py that filters Entries using the active language

class LocaleEntryManager(models.Manager, ActiveAwareContentManagerMixin):
    def get_query_set(self):
        return TransformQuerySet(self.model, using=self._db)
    def featured(self):
        return self.active().filter(is_featured=True)
LocaleEntryManager.add_to_active_filters(
    Q(is_active=True),
    key='cleared')
LocaleEntryManager.add_to_active_filters(
    lambda queryset: queryset.filter(published_on__lte=now),
    key='published_on_past')

LocaleEntryManager.add_to_active_filters(
    lambda queryset:queryset.filter(language__startswith=short_language_code()),
    key='active_language')

In the Entry class i added a further variable (look at the last line):

    class Meta:
        get_latest_by = 'published_on'
        ordering = ['-published_on']
        verbose_name = _('entry')
        verbose_name_plural = _('entries')
    objects = EntryManager()
    locale_objects = LocaleEntryManager() 

The final step is to replace the object variable with locale_object in feeds.py and views.py.

In my setup, I want to have everything included in /sitemap.xml and everything else seperated by language. This works like a charm for me now:

http://www.simme.org/sitemap.xml

http://www.simme.org/de/blog/feed

http://www.simme.org/en/blog/feed

In addition, this solution makes it pretty simple to add a few parameters to settings.py in order to decide which variable should be used for what/where.

What do you think?

Another possibility would be to check where the request comes from and then add the "active_language_filter" if necessary / wanted e.g. for the feed and the entry_archive.

[sbaechler]

Those are the two approaches. The issue here is that the language field is not present if the translation extension is not used. And since this is a class based view, you could define locale_objects as a mixin. And if you do a mixin anyway, you could just do the filtering there.

[sbaechler]

I added a test suite so it's possible to test the translation implementation once we agree on one.

I slightly modified Matthias' approach of extending the get_queryset method. But in a separate Mixin instead of the main Elephantblog Mixin.

We could discuss if it's better to simply add the Mixin to the views or dynamically add it from the URLs.

https://github.com/sbaechler/feincms-elephantblog/commit/1039c5a09e524aee500a5a0e13a0c36e54710d72

[simonkern]

Another factor that should be considered is the following:

When using feincms's feincms_page model, you can spit out links that lead directly to the translated page. E.g. by using:

{% feincms_languagelinks for feincms_page as links all,excludecurrent %}
{% for key, name, link, translation in links %}
   {% if link %}


{% trans "Switch language" %}: {{ key }}{% endif %}
{% endfor %}

When you try to do the same with elephantblog, it gives you a wrong absolute url, because it does not take (as far as u used it) the language switch at the root level of your page into it's consideration. So you can only provide a link to the translated's blog root level, but not yet to a specific article.

For example you have a blog entry in german available at:

http://www.excample.org/de/blog/2012/10/21/test-eintrag/

It's translation link will point to:

http://www.example.org/de/blog/2012/10/21/test-entry/

Although it should be:

http://www.example.org/en/blog/2012/10/21/test-entry/

There should be a "language-switch" awareness somehow.

A simple workaround for now might be to replace the string of the actual language in the url and replace it by the language returned in the key variable.

[sbaechler]

It actually goes further. The page module should know in which languages the entries are available and only show those languages as options. Currently the selectable languages in the parent page are the translations of that page.

[sbaechler]

I have noticed that people start forking elephantblog to implement their own solution for the language problem. We should expedite finding a common solution.

[simonkern]

Here is a suggestion:

Since elephantblog is most likely used together with feincms, it might be a good idea to add a language parameter to feincms's Apllication Content (e.g. english, german or both - depending on available languages in settings.py). This parameter could be used to determine which entries (depending on their language) should be served under which path.

This might also solve the multiple occurences of entries in the rss-feed, when elephantblog is added to every language subtree as Application Content.

[matthiask]

Fixed:

https://github.com/feincms/feincms-elephantblog/blob/master/docs/installation.rst#integrating-standalone

[saboter]

Hi @matthiask,

posting again the same comment (deleted accidentally the old one).

Any hints how to make this new setting work with elephantblog integrated as application content? Adding this snippet didn't helped:

from elephantblog.urls import elephantblog_patterns
urlpatterns += patterns('',
    url(r'^blog/', include(elephantblog_patterns(
        list_kwargs={'only_active_language': True},
        ))),
)

Blog entries are not filtered by active language. I tried to set/hardcode the list_kwargs={'only_active_language': True} directly into elephantblog_patterns method declaration as a test. It worked after that of course.

[matthiask]

@saboter Where did you put this elephantblog_patterns invocation? And which file did you reference when registering the ApplicationContent?

If you put this code into myapp.blog_urls, you'll also have to use myapp.blog_urls when registering the blog app with ApplicationContent.

[saboter]

@matthiask you are right of course. I messed up the urls setup, shame on me. Thank you for assistance.

[simonkern]

as soon as i register the translation extension to Entry (as shown in the docs) I get the following error: There are duplicate field(s) in EntryAdmin.fieldsets

Traceback:
File "/srv/blog/.env/local/lib/python2.7/site-packages/django/core/handlers/base.py" in get_response
  92.                     response = middleware_method(request)
File "/srv/blog/.env/local/lib/python2.7/site-packages/django/middleware/locale.py" in process_request
  21.         check_path = self.is_language_prefix_patterns_used()
File "/srv/blog/.env/local/lib/python2.7/site-packages/django/middleware/locale.py" in is_language_prefix_patterns_used
  56.         for url_pattern in get_resolver(None).url_patterns:
File "/srv/blog/.env/local/lib/python2.7/site-packages/django/core/urlresolvers.py" in url_patterns
  347.         patterns = getattr(self.urlconf_module, "urlpatterns", self.urlconf_module)
File "/srv/blog/.env/local/lib/python2.7/site-packages/django/core/urlresolvers.py" in urlconf_module
  342.             self._urlconf_module = import_module(self.urlconf_name)
File "/srv/blog/.env/local/lib/python2.7/site-packages/django/utils/importlib.py" in import_module
  35.     __import__(name)
File "/srv/blog/blog/urls.py" in 
  8. admin.autodiscover()
File "/srv/blog/.env/local/lib/python2.7/site-packages/django/contrib/admin/__init__.py" in autodiscover
  29.             import_module('%s.admin' % app)
File "/srv/blog/.env/local/lib/python2.7/site-packages/django/utils/importlib.py" in import_module
  35.     __import__(name)
File "/srv/blog/.env/src/elephantblog/elephantblog/admin.py" in 
  28. admin.site.register(models.Entry, models.EntryAdmin)
File "/srv/blog/.env/local/lib/python2.7/site-packages/django/contrib/admin/sites.py" in register
  98.                 validate(admin_class, model)
File "/srv/blog/.env/local/lib/python2.7/site-packages/django/contrib/admin/validation.py" in validate
  25.     validate_base(cls, model)
File "/srv/blog/.env/local/lib/python2.7/site-packages/django/contrib/admin/validation.py" in validate_base
  300.             raise ImproperlyConfigured('There are duplicate field(s) in %s.fieldsets' % cls.__name__)
Exception Type: ImproperlyConfigured at /
Exception Value: There are duplicate field(s) in EntryAdmin.fieldsets