A Django package that provides auto indexing and searching capabilities for Django model instances using RediSearch.
- Management Command to create, update and populate the RediSearch Index.
- Auto Index on Model object Create, Update and Delete.
- Auto Index on Related Model object Add, Update, Remove and Delete.
- Easy to create Document classes (Uses Django Model Form Class like structure).
- Index nested models (e.g:
OneToOneField
,ForeignKey
andManyToManyField
). - Search documents using
redis-om
. - Search Result Pagination.
- Search Result Sorting.
- RediSearch Result to Django QuerySet.
- Faceted Search.
- Python: 3.7, 3.8, 3.9, 3.10
- Django: 3.2, 4.0, 4.1
- redis-om: >= 0.0.27
The latest version of Redis is available from Redis.io. You can also install Redis with your operating system's package manager.
redis-search-django
relies on the RediSearch and RedisJSON Redis modules to support rich queries and embedded models.
You need these Redis modules to use redis-search-django
.
The easiest way to run these Redis modules during local development is to use the redis-stack Docker image.
There is a docker-compose.yaml
file provided in the project's root directory.
This file will run Redis with RedisJSON and RediSearch modules during development.
Run the following command to start the Redis container:
docker compose up -d
There is an example project available at Example Project.
pip install redis-search-django
Then add redis_search_django
to your INSTALLED_APPS
:
INSTALLED_APPS = [
...
'redis_search_django',
]
There are 3 types of documents class available:
- JsonDocument: This uses
RedisJSON
to store the document. If you want to use Embedded Documents (Required ForOneToOneField
,ForeignKey
andManyToManyField
) then useJsonDocument
. - EmbeddedJsonDocument: If the document will be embedded inside another document class then use this. Embedded Json Documents are used for
OneToOneField
,ForeignKey
andManyToManyField
or any types of nested documents. - HashDocument: This uses
RedisHash
to store the documents. It can not be used for nested documents.
You need to inherit from The Base Document Classes mentioned above to build a document class.
1. For Django Model:
# models.py
from django.db import models
class Category(models.Model):
name = models.CharField(max_length=30)
slug = models.SlugField(max_length=30)
def __str__(self) -> str:
return self.name
2. You can create a document class like this:
Note: Document classes must be stored in documents.py
file.
# documents.py
from redis_search_django.documents import JsonDocument
from .models import Category
class CategoryDocument(JsonDocument):
class Django:
model = Category
fields = ["name", "slug"]
3. Run Index Django Management Command to create the index on Redis:
python manage.py index
Note: This will also populate the index with existing data from the database
Now category objects will be indexed on create/update/delete.
1. For Django Models:
# models.py
from django.db import models
class Tag(models.Model):
name = models.CharField(max_length=30)
def __str__(self) -> str:
return self.name
class Vendor(models.Model):
name = models.CharField(max_length=30)
email = models.EmailField()
establishment_date = models.DateField()
def __str__(self) -> str:
return self.name
class Product(models.Model):
name = models.CharField(max_length=256)
description = models.TextField(blank=True)
vendor = models.OneToOneField(Vendor, on_delete=models.CASCADE)
tags = models.ManyToManyField(Tag, blank=True)
price = models.DecimalField(max_digits=6, decimal_places=2)
def __str__(self) -> str:
return self.name
2. You can create a document classes like this:
Note: Document classes must be stored in documents.py
file.
# documents.py
from typing import List
from django.db import models
from redis_om import Field
from redis_search_django.documents import EmbeddedJsonDocument, JsonDocument
from .models import Product, Tag, Vendor
class TagDocument(EmbeddedJsonDocument):
custom_field: str = Field(index=True, full_text_search=True)
class Django:
model = Tag
# Model Fields
fields = ["name"]
@classmethod
def prepare_custom_field(cls, obj):
return "CUSTOM FIELD VALUE"
class VendorDocument(EmbeddedJsonDocument):
class Django:
model = Vendor
# Model Fields
fields = ["name", "establishment_date"]
class ProductDocument(JsonDocument):
# OnetoOneField, with null=False
vendor: VendorDocument
# ManyToManyField
tags: List[TagDocument]
class Django:
model = Product
# Model Fields
fields = ["name", "description", "price"]
# Related Model Options
related_models = {
Vendor: {
"related_name": "product",
"many": False,
},
Tag: {
"related_name": "product_set",
"many": True,
},
}
@classmethod
def get_queryset(cls) -> models.QuerySet:
"""Override Queryset to filter out available products."""
return super().get_queryset().filter(available=True)
@classmethod
def prepare_name(cls, obj):
"""Use this to update field value."""
return obj.name.upper()
Note:
- You can not inherit from
HashDocument
for documents that include nested fields. - You need to inherit from
EmbeddedJsonDocument
for document classes that will be embedded inside another document class. - You need to explicitly add
OneToOneField
,ForeignKey
orManyToManyField
(e.g:tags: List[TagDocument]
) with an embedded document class if you want to index them. you can not add it in theDjango.fields
option. - For
related_models
option, you need to specify the fieldsrelated_name
and if it is aManyToManyField
or aForeignKey
Field then specify"many": True
. related_models
will be used when a related object is saved that contributes to the document.- You can define
prepare_{field_name}
method to update the value of a field before indexing. - If it is a custom field (not a model field) you must define a
prepare_{field_name}
method that returns the value of the field. - You can override
get_queryset
method to provide more filtering. This will be used while indexing a queryset. - Field names must match model field names or define a
prepare_{field_name}
method.
3. Run Index Django Management Command to create the index on Redis:
python manage.py index
Note: This will also populate the index with existing data from the database
This package comes with index
management command that can be used to index all the model instances to Redis index if it has a Document class defined.
Note: Make sure that Redis is running before running the command.
Run the following command to index all models that have Document classes defined:
python manage.py index
You can use --migrate-only
option to only update the index schema.
python manage.py index --migrate-only
You can use --models
to specify which models to index (models must have a Document class defined to be indexed).
python manage.py index --models app_name.ModelName app_name2.ModelName2
You can use the redis_search_django.mixin.RediSearchListViewMixin
with a Django Generic View to search for documents.
RediSearchPaginator
which helps paginate ReadiSearch
results is also added to this mixin.
# views.py
from django.utils.functional import cached_property
from django.views.generic import ListView
from redis.commands.search import reducers
from redis_search_django.mixins import RediSearchListViewMixin
from .documents import ProductDocument
from .models import Product
class SearchView(RediSearchListViewMixin, ListView):
paginate_by = 20
model = Product
template_name = "core/search.html"
document_class = ProductDocument
@cached_property
def search_query_expression(self):
query = self.request.GET.get("query")
query_expression = None
if query:
query_expression = (
self.document_class.name % query
| self.document_class.description % query
)
return query_expression
@cached_property
def sort_by(self):
return self.request.GET.get("sort")
def facets(self):
if self.search_query_expression:
request = self.document_class.build_aggregate_request(
self.search_query_expression
)
else:
request = self.document_class.build_aggregate_request()
result = self.document_class.aggregate(
request.group_by(
["@tags_name"],
reducers.count().alias("count"),
)
)
return result
This package uses redis-om
to search for documents.
from .documents import ProductDocument
categories = ["category1", "category2"]
tags = ["tag1", "tag2"]
# Search For Products That Match The Search Query (name or description)
query_expression = (
ProductDocument.name % "Some search query"
| ProductDocument.description % "Some search query"
)
# Search For Products That Match The Price Range
query_expression = (
ProductDocument.price >= float(10) & ProductDocument.price <= float(100)
)
# Search for Products that include following Categories
query_expression = ProductDocument.category.name << ["category1", "category2"]
# Search for Products that include following Tags
query_expression = ProductDocument.tags.name << ["tag1", "tag2"]
# Query expression can be passed on the `find` method
result = ProductDocument.find(query_expression).sort_by("-price").execute()
For more details checkout redis-om docs
redis-om
does not support faceted search (RediSearch Aggregation). So this package uses redis-py
to do faceted search.
from redis.commands.search import reducers
from .documents import ProductDocument
query_expression = (
ProductDocument.name % "Some search query"
| ProductDocument.description % "Some search query"
)
# First we need to build the aggregation request
request1 = ProductDocument.build_aggregate_request(query_expression)
request2 = ProductDocument.build_aggregate_request(query_expression)
# Get the number of products for each category
ProductDocument.aggregate(
request1.group_by(
["@category_name"],
reducers.count().alias("count"),
)
)
# >> [{"category_name": "Shoes", "count": "112"}, {"category_name": "Cloths", "count": "200"}]
# Get the number of products for each tag
ProductDocument.aggregate(
request2.group_by(
["@tags_name"],
reducers.count().alias("count"),
)
)
# >> [{"tags_name": "Blue", "count": "14"}, {"tags_name": "Small", "count": "57"}]
For more details checkout redis-py docs and RediSearch Aggregation docs
REDIS_OM_URL
(Default:redis://localhost:6379
): This environment variable follows theredis-py
URL format. If you are using external redis server You need to set this variable with the URL of the redis server following this pattern:redis://[[username]:[password]]@[host]:[post]/[database number]
Example: redis://redis_user:password@some.other.part.cloud.redislabs.com:6379/0
For more details checkout redis-om docs
You can add these options on the Django
class of each Document class:
# documents.py
from redis_search_django.documents import JsonDocument
from .models import Category, Product, Tag, Vendor
class ProductDocument(JsonDocument):
class Django:
model = Product
fields = ["name", "description", "price", "created_at"]
select_related_fields = ["vendor", "category"]
prefetch_related_fields = ["tags"]
auto_index = True
related_models = {
Vendor: {
"related_name": "product",
"many": False,
},
Category: {
"related_name": "product_set",
"many": True,
},
Tag: {
"related_name": "product_set",
"many": True,
},
}
model
(Required): Django Model class to index.auto_index
(Default:True
, Optional): If True, the model instances will be indexed on create/update/delete.fields
(Default:[]
, Optional): List of model fields to index. (Do not addOneToOneField
,ForeignKey
orManyToManyField
here. These need to be explicitly added to the Document class usingEmbeddedJsonDocument
.)select_related_fields
(Default:[]
, Optional): List of fields to use onqueryset.select_related()
.prefetch_related_fields
(Default:[]
, Optional): List of fields to use onqueryset.prefetch_related()
.related_models
(Default:{}
, Optional): Dictionary of related models. You need to specify the fieldsrelated_name
and if it is aManyToManyField
or aForeignKey
Field then specify"many": True
. These are used to update the document data if any of the related model instances are updated.related_models
will be used when a related object is saved/added/removed/deleted that contributes to the document.
For redis-om
specific options checkout redis-om docs
You can add these options to your Django settings.py
File:
REDIS_SEARCH_AUTO_INDEX
(Default:True
): Enable or Disable Auto Index when model instance is created/updated/deleted for all document classes.
The code in this project is released under the MIT License.