jorgecarleitao/django-sphinxql

Django-SphinxQL implements Sphinx search for Django, thanks for checking it out.

Django is a Web framework for building websites with relational databases; Sphinx is a search engine designed for relational databases. Django-SphinxQL defines an ORM for using Sphinx in Django. As corollary, it allows you to implement full text search with Sphinx in your Django website.

Specifically, this API allows you to:

Configure Sphinx with Python.
Index Django models in Sphinx.
Execute Sphinx queries (SphinxQL) using Django-like expressions and have the results as Django models.

Django-SphinxQL requires:

Python 3
pymysql
Django (>=1.8)
Sphinx
A backend (pymysql or psycopg2)

Our build matrix in Travis has 8 builds:

Python 3.4
Django 1.8 and 1.10
Sphinx 2.2.10 and 2.2.11
mysql and postgres backends

For more details, you can check the directory tests and .travis.yml.

To run the tests locally, use:

PYTHONPATH=..:$PYTHONPATH django-admin.py test --settings=tests.settings_test tests

The next sections present a minimal setup to use this package. The full documentation is available here.

Installation

Django-SphinxQL has no requirements besides Django and Sphinx. To install Sphinx, use:

export VERSION=2.2.10
wget http://sphinxsearch.com/files/sphinx-$VERSION-release.tar.gz
tar -xf sphinx-$VERSION-release.tar.gz
cd sphinx-$VERSION-release
./configure --prefix=$HOME --with-pgsql
make
make install

To install Django-SphinxQL, use:

pip install git+https://github.com/jorgecarleitao/django-sphinxql.git

Minimal configuration

Django-SphinxQL requires a directory to store its database and be registered as installed app (it doesn't contain Django models):

add sphinxql to the INSTALLED_APPS;

add INDEXES to settings:

 INDEXES = {
     'path': os.path.join(BASE_DIR, '_index'),  # also do `mkdir _index`.
     'sphinx_path': BASE_DIR
 }

path is where Sphinx database is going to be created
sphinx_path is the directory that will contain Sphinx-specific files.

Index your models

Assume you have a model Document with a summary, a text and a number that you want to index. To index it, create a file indexes.py in your app with:

from sphinxql import indexes, fields
from myapp import models

class DocumentIndex(indexes.Index):
    my_summary = fields.Text(model_attr='summary')
    my_text = fields.Text(model_attr='text')
    my_number = fields.Integer(model_attr='number')

    class Meta:
        model = models.Document

model_attr can be either a string with lookups or an F expression. E.g. type_name = fields.Text(model_attr='type__name') will index the name of the ForeignKey type of your model, while

type_name = fields.Text(model_attr=Concat('type__name', Value(' '),
                                          'my_text',
                                          output_field=CharField()))

indexes the concatenation of two fields (see also Django documentation). In principle the index fields accept any Django expression Django annotate accepts.

To index your indexes, run:

python manage.py index_sphinx

At this moment you may notice that some files will be created in settings.INDEXES['path']: Sphinx database is populated.

Then, start Sphinx daemon (only has to be started once):

python manage.py start_sphinx

(for the sake of reversibility, to stop Sphinx use python manage.py stop_sphinx)

Search your indexes

Django-SphinxQL defines a subclass of Django QuerySet's, that interfaces with all Sphinx-related operations. SearchQuerySet only adds functionality: if you don't use Sphinx-related, it is a QuerySet.

Sphinx has a dedicated syntax for text search that Django-SphinxQL also accepts:

>>> q = SearchQuerySet(DocumentIndex).search('@my_text toys for babies')

This particular query returns Documents restricted to the ones where "toys for babies" match in field my_text, ordered by the most relevant match. Once you perform it, it does the following:

hit Sphinx database and convert results to DocumentIndex instances;
hit Django database to retrieve the respective Document instances;
annotate Document instances with the respective DocumentIndex instances (in attribute search_result)
returns the Document instances.

Step 2. is done using .filter(pk__in=[...]). The results are ordered by relevance because there was no specific call of order_by: if you set any ordering in Django Query, it uses Django ordering (i.e. it overrides the default ordering but not an explicit ordering). See docs for detailed information.

Known limitations

Null values are considered empty strings or 0 (constraint on Sphinx engine)
Only supports dates and times since 1970 (constraint on Sphinx engine)
Most Sphinx functionality is not implemented, notably real time indexes.

Final note

You should check if Django-Haystack suits your needs.

Django-SphinxQL is useful when you can index your data on a time-scale different from "real time". It should be much faster in indexing, and it should have lower memory requirements.

jorgecarleitao / django-sphinxql