Question

I have two models like this:

class ClassA(models.Model):
    ida = models.AutoField(primary_key=True)
    classb = models.ForeignKey(ClassB)
    name = models.CharField(max_length=765)

class ClassB(models.Model):
    idb = models.AutoField(primary_key=True)
    name = models.CharField(max_length=765)

I want to show a list of all my ClassA objects grouped under a ClassB heading. This is the view I have:

def show_class_a_objects(request):
    class_b_objects = ClassB.objects.all().order_by('name')
    class_a_objects = ClassA.objects.all().order_by('name')
    return render_to_response('show_objects.html', {
        'class_a_objects': class_a_objects, 'class_a_objects': class_a_objects, 
    })

My show_objects.html file looks like this:

{% extends "base_show.html" %}
{% block content %}
</p>
<table>
<th align="left">Name</th>
{% for b in class_b_objects %}
    <tr>
    <td>{{b.name}}</td>
    </tr>
    {% for a in class_a_objects %}
        {% ifequal a.classb b %}
        <tr>
        <td><a href="{{ a.get_absolute_url }}">{{a.name}}</a></td>
        </tr>
        {% endifequal %}
    {% endfor %}
{% endfor %}
</table>
</br>
{% endblock %}

Is there a better way of doing this instead of using the two nested for loops?

Answer 1

def show_class_a_objects(request):
    class_b_objects = ClassB.objects.all().order_by('name')
    class_a_objects = ClassA.objects.all().order_by('name')

You don’t need to query the ClassA objects like this to achieve the grouping you want. When you create a ForeignKey on your ClassA -> ClassB, your ClassB will contain a relation that will give you back all ClassA objects which have ClassB as a ForeignKey:

view

def show_class_a_objects(request):
    class_b_objects = ClassB.objects.all().order_by('name')
    return render_to_response('show_objects.html', {
        'class_b_objects': class_b_objects, 
    })

template

{% for b in class_b_objects %}
    <tr>
    <td>{{b.name}}</td>
    </tr>
    {% for a in b.classa_set.all %}
        <tr>
        <td><a href="{{ a.get_absolute_url }}">{{a.name}}</a></td>
        </tr>
    {% endfor %}
{% endfor %}

You still need a nested for loop, but you no longer have to loop over the entire range of every single other class and do comparisons to find the right ones.

If you want a bit more control, you should move this into your view again and prebuild the dict:

view

def show_class_a_objects(request):
    class_b = ClassB.objects.all().order_by('name')
    class_b_objects = [(b, b.classa_set.all().order_by('name')) for b in class_b]
    return render_to_response('show_objects.html', {
        'class_b_objects': class_b_objects
    })

template

{% for b, a_list in class_b_objects %}
    <tr>
    <td>{{b.name}}</td>
    </tr>
    {% for a in a_list %}
        <tr>
        <td><a href="{{ a.get_absolute_url }}">{{a.name}}</a></td>
        </tr>
    {% endfor %}
{% endfor %}

Update: Reducing number of queries

As per the comments, I wanted to address the issue of increasing the amount of database queries from your original 2, to the amount of 1 + num_of_classb_objects. This is because while the original ClassB query is 1, each of the b.classa_set.all() relation calls is another query to fetch the related ClassA instances.

If you want to try and test an alternative, you can check out this blog post on efficient reverse lookups

Example (adapted from link to your model names):

qs = ClassB.objects.all()
obj_dict = dict([(obj.id, obj) for obj in qs])
objects = ClassA.objects.filter(classb__in=qs)
relation_dict = {}
for obj in objects:
    relation_dict.setdefault(obj.classb_id, []).append(obj)
for id, related_items in relation_dict.items():
    obj_dict[id]._related_items = related_items

Also, if you are running django >= 1.4, this functionality is supposedly official in the form of prefetch_related.

I have not tested this, but I wanted to present it as something for you to look into if n+1 queries are not acceptable to you.