A Multiple Model Django Search Engine
By Justin
The Django ORM makes it simple to search. In this one, we'll create an internal search that can be rather useful.
Getting lost on this guide? Perhaps watching any of the Try Django series would help.
Model Examples
python
# blog.models
class Post(models.Model):
user = models.ForeignKey(settings.AUTH_USER_MODEL, on_delete=models.CASCADE)
title = models.CharField(max_length=120)
description = models.TextField(null=True, blank=True)
slug = models.SlugField(blank=True, unique=True)
publish_date = models.DateTimeField(auto_now_add=False, auto_now=False, null=True, blank=True)
timestamp = models.DateTimeField(auto_now_add=True)
# courses.models
class Lesson(models.Model):
title = models.CharField(max_length=120)
description = models.TextField(null=True, blank=True)
slug = models.SlugField(blank=True, unique=True)
featured = models.BooleanField(default=False)
publish_date = models.DateTimeField(auto_now_add=False, auto_now=False, null=True, blank=True)
# profiles.models
class Profile(models.Model):
user = models.OneToOneField(settings.AUTH_USER_MODEL, on_delete=models.CASCADE)
title = models.CharField(max_length=120)
description = models.TextField(null=True, blank=True)
slug = models.SlugField(blank=True, unique=True)
timestamp = models.DateTimeField(auto_now_add=True)
Simple Search (aka Lookup)
>>> from django.utils import timezone
>>> from blog.models import Post
>>> qs = Post.objects.filter(publish_date__lte=timezone.now(), title__icontains="Django")
>>> from django.utils import timezone
>>> from courses.models import Lesson
>>> qs = Lesson.objects.filter(publish_date__lte=timezone.now(), featured=True)
>>> from profiles.models import Profile
>>> obj = Profile.objects.get(user__id=1)
Advancing the Query Lookup
Here's a few more queries that are a little more robut because of Q Lookups. You might want to learn more about lookups in the Products Component or specifically in Understanding Lookups.
>>> from django.db.models import Q
>>> from django.utils import timezone
>>> from blog.models import Post
>>> query = "Django"
>>> or_lookup = (Q(title__icontains=query) | Q(description__icontains=query))
>>> print(or_lookup)
(OR: ('title__icontains', 'Django'), ('description__icontains', 'Django'))
>>> and_lookup = (Q(title__icontains=query) & Q(description__icontains=query))
>>> print(and_lookup)
(<Q: (AND: ('title__icontains', 'Django'))>, <Q: (AND: ('description__icontains', 'Django'))>)
>>> qs_and = Post.objects.filter(publish_date__lte=timezone.now()).filter(and_lookup)
>>> print(qs_and)
>>> qs_or = Post.objects.filter(publish_date__lte=timezone.now()).filter(or_lookup)
>>> print(qs_or)
Now that we see how easy Q Lookups are, we should implement a method to our model manager to handle any given query. Below is an example on the Post model, I'll leave it to you to implement your own.
Update Model Manager to include a Search Method
python
# blog.models
from django.db.models import Q
class PostManager(models.Manager):
def search(self, query=None):
qs = self.get_queryset()
if query is not None:
or_lookup = (Q(title__icontains=query) |
Q(description__icontains=query)|
Q(slug__icontains=query)
)
qs = qs.filter(or_lookup).distinct() # distinct() is often necessary with Q lookups
return qs
class Post(models.Model):
user = models.ForeignKey(settings.AUTH_USER_MODEL)
title = models.CharField(max_length=120)
description = models.TextField(null=True, blank=True)
slug = models.SlugField(blank=True, unique=True)
publish_date = models.DateTimeField(auto_now_add=False, auto_now=False, null=True, blank=True)
timestamp = models.DateTimeField(auto_now_add=True)
objects = PostManager()
Now, let's implement this concept into a view.
Create the search view
python
# search.views.py
from itertools import chain
from django.views.generic import ListView
from blog.models import Post
from courses.models import Lesson
from profiles.models import Profile
class SearchView(ListView):
template_name = 'search/view.html'
paginate_by = 20
count = 0
def get_context_data(self, *args, **kwargs):
context = super().get_context_data(*args, **kwargs)
context['count'] = self.count or 0
context['query'] = self.request.GET.get('q')
return context
def get_queryset(self):
request = self.request
query = request.GET.get('q', None)
if query is not None:
blog_results = Post.objects.search(query)
lesson_results = Lesson.objects.search(query)
profile_results = Profile.objects.search(query)
# combine querysets
queryset_chain = chain(
blog_results,
lesson_results,
profile_results
)
qs = sorted(queryset_chain,
key=lambda instance: instance.pk,
reverse=True)
self.count = len(qs) # since qs is actually a list
return qs
return Post.objects.none() # just an empty queryset as default
Create a new template tag to get the class name
python
# search.templatetags.class_name.py
from django import template
register = template.Library()
@register.filter()
def class_name(value):
return value.__class__.__name__
Create the view template
search/view.html{% extends "base.html" %}
{% load class_name %}
{% block content %}
<div class='row title-row my-5'>
<div class='col-12 py-0'>
<h3 class='my-0 py-0'>{{ count }} results for <b>{{ query }}</b></h3>
</div>
</div>
{% for object in object_list %}
{% with object|class_name as klass %}
{% if klass == 'Post' %}
<div class='row'>
<div class='col-12'>
Blog post: <a href='{{ object.get_absolute_url }}'>{{ object.title }}</a>
</div>
</div>
{% elif klass == 'Lesson' %}
<div class='row'>
<div class='col-12'>
Lesson Item: <a href='{{ object.get_absolute_url }}'>{{ object.title }}</a>
</div>
</div>
{% elif klass == 'Profile' %}
<div class='row'>
<div class='col-12'>
Lesson Item: <a href='{{ object.get_absolute_url }}'>{{ object.title }}</a>
</div>
</div>
{% else %}
<div class='row'>
<div class='col-12 col-lg-8 offset-lg-4'>
<a href='{{ object.get_absolute_url }}'>{{ object }} | {{ object|class_name }}</a>
</div>
</div>
{% endif %}
{% endwith %}
{% empty %}
<div class='row'>
<div class='col-12 col-md-6 mx-auto my-5 py-5'>
<form method='GET' class='' action='.'>
<div class="input-group form-group-no-border mx-auto" style="margin-bottom: 0px; font-size: 32px;">
<span class="input-group-addon cfe-nav" style='color:#000'>
<i class="fa fa-search" aria-hidden="true"></i>
</span>
<input type="text" name="q" data-toggle="popover" data-placement="bottom" data-content="Press enter to search" class="form-control cfe-nav mt-0 py-3" placeholder="Search..." value="" style="" data-original-title="" title="" autofocus="autofocus">
</div>
</form>
</div>
</div>
{% endfor %}
{% endblock content %}
Even more advanced searching...
See from here, we can advance our search even more. This part qs = sorted(queryset_chain, key=lambda instance: instance.pk, reverse=True) shows us we can reorder our responses based on an aribratry field, or even an instance method.
That means we can create a method to calculate "rank" for any given model and any given model instance. This rank, can be based off analytics like link clicks, page views, link social shares, or any other rank parameter we might like. Thus creating a better experience for your searches.
Do you want to see a "page rank" search implemented? Let us know in the comments.