patrickomatic / django-full-text-search

A simple database-agnostic full text search for Django

django-full-text-search

django-full-text-search provides a simple, database-agnostic full text search for Django models. By extending Django's model manager functionality, searching and indexing of documents can be added to any Django model. django-full-text-search also provides pluggable stemming implementations and includes (and defaults to) an implementation of the Porter stemming algorithm.

Overview

Once installed, searching documents is as easy as:

1. Registering a manager for the model you would like to make searchable:

from ftsearch.managers import SearchableManager

class MyModel(models.Model):

...

objects = SearchableManager()

2. Define a get_text_only() method on the model. This method should return a single string containing all text to be searched:

class MyModel(models.Model):
   name = models.CharField()
   body = models.CharField()
   ...
   objects = SearchableManager()

   def get_text_only(self):
      return ' '.join(self.name, self.body)

3. Index the documents when they are created or modified by using the manager methods. This only needs to be done once:

MyModel.objects.add_to_index(my_model_instance)

4. Now search the indexed documents:

   MyModel.objects.search('my search terms')

Installation

1. Download and install django-full-text-search using Distutils:

$ sudo python setup.py install

2. Add the ftsearch app to INSTALLED_APPS in your settings.py

Configuration

By default django-full-text-search comes with pretty reasonable defaults and doesn't require any additional configuration. However, if you would like to override some defaults, you can use the following configuration options:

• SEARCH_WEIGHTS - A list of 2-value tuples which specify the weights to be used w hen ranking results. if you would like to add application-specific weighting rules, you can do so by overriding SEARCH_WEIGHTS. By default it is defined as:

  settings.SEARCH_WEIGHTS = (
                  (1.0, ftsearch.weights.frequency_score),
                  (1.0, ftsearch.weights.location_score),
                  (1.0, ftsearch.wieghts.distance_score),
  )
  

• SEARCH_STEMMER - A custom stemming implementation. It must be a class which has a stem() method.

• SEARCH_WORD_SPLIT_REGEX - The regex that is used to split text. By default it is re.compile(r'\W*').

Limitations

django-full-text-search is mostly suitable for small datasets. If you need to index and search millions of documents, it may not be for you.