ikollipara / django-meili

Django integration with Meilisearch

Django-Meili

A package to integrate Meilisearch with Django in a seamless way.

Usage

Set the following variables in settings.py.

MEILISEARCH = {
   "HOST": "localhost",
   "PORT": 7700,
   "HTTPS": False,
   "MASTER_KEY": "...",
   "SYNC": False, # Should Meilisearch action resolve before continuing
}

Register the app in INSTALLED_APPS

INSTALLED_APPS = [
    ...,
    "django_meili",
    ...,
]

Just subclass django_meili.models.IndexMixin

from django_meili.models import IndexMixin

class Post(IndexMixin, models.Model):
    id = models.UUIDField()
    title = models.CharField(max_length=255)
    body = models.TextField()

    # Attributes to handle in Meilisearch
    displayed_fields = ("title", "body")
    searchable_fields = ("title", "body")

    def __str__(self):
        return self.title

Then when you are ready to search this model, just use the built-in "queryset".

Post.objects.create(title="Hello", body="World")
posts = Post.meilisearch.search("hello") # Returns a Django Queryset of results
print(posts.first()) # Hello

API

django_meili.models.IndexQuerySet

This is a custom queryset built to mimic Django's queryset. Currently does not support Q objects, but that is a plan for the future.

Methods:

Name	Description	Example
getitem	Use slices to set limit and offset. The default is 20 and 0 respectively.	Post.meilisearch[:10]
count	Return the total number of documents within the index. Does not reflect the count for the search query	Post.meilisearch.count()
order_by	Takes a Django order_by parameter and sets a sort value based on that. Can take multiple sorts	Post.meilisearch.order_by("-likes")
filter	Takes a Django filter query. Supports: lte, gte, lt, gt, exact, in, range, isnull, exists	Post.meilisearch.filter(title__exact="Hello World")
matching_strategy	Set the default strategy. Defaults to last	Post.meilisearch.matching_strategy('all')
attributes_to_search_on	Choose what fields to search on, defaults to all.	Post.meilisearch.attributes_to_search_on("title", "body")
search	Executes the actual search. Takes an optional query and returns a Django Queryset of results	Post.meilisearch.search()

django_meili.models.IndexMixin

Subclass Parameters:

Name	Type	Description
index_name	Optional[str]	The name of the index to generate. Defaults to __name__ attribute.
primary_key	Optional[str]	The primary key of the index. Defaults to pk.

Attributes:

Name	Type	Description
displayed_fields	tuple[str]	The fields to display. By default uses all.
searchable_fields	tuple[str]	The fields to search. By default uses all.
filterable_fields	tuple[str]	The fields to filter. By default uses none.
meilisearch	django_meili.models.IndexQuerySet	The queryset used for filtering and searching

---

Copyright 2024 Ian Kollipara <[email protected]>

django-meili's Issues

Potential error point

This code could potentially error. The reason is create_index is asyncronous in Meilisearch so if your code tries to update the settings before the index gets created it will error. You are unlikely to see this in development because there won't be many tasks and therefore the index should get created quickly.

One solution would be to wait for the index creation to complete. The thing to be aware of here is the program will be blocked while it waits. Most of the time this will be fast and not noticable, but it's something to be aware of. If you want me to do a PR to add this let me know and I can.

If you decide to leave things as they are it will be a race condition kind of bug. It will most likely be rare that you see it, so just be aware that if you see an error here that you can't explain or easily reproduce this could be the issue.

django-meili/src/django_meili/models.py

Lines 43 to 57 in 5319ae2

	if index_name not in _current_indices:
	client.create_index(
	cls.__name__ if not index_name else index_name,
	{"primaryKey": primary_key},
	)
	if cls.displayed_fields:
	client.index(cls.__name__).update_displayed_attributes(cls.displayed_fields)
	if cls.searchable_fields:
	client.index(cls.__name__).update_searchable_attributes(
	cls.searchable_fields
	)
	if cls.filterable_fields:
	client.index(cls.__name__).update_filterable_attributes(
	cls.filterable_fields
	)