Skip to content

Latest commit

 

History

History
102 lines (80 loc) · 3.31 KB

README.md

File metadata and controls

102 lines (80 loc) · 3.31 KB
Raw

If you are using Djapian please tell us about your project in reply to this post

Use this package to allow full-text search in your Django project.

Versions compatibility matrix:

Djapian Django Xapian
<= 2.2.4 1.0 1.0.2
<= 2.3.1 1.1 1.0.7
>= 2.4 1.2.1 1.2

Notice: there is an old issue with Xapian (< 1.0.13) in mod_python environment. So be careful.

Notice: with 2.2.2 release has been introduced database schema backward-incompatible bug fix - Change model has switched its object_id field type from integer to string.

###Features

Most of this features provided by Xapian itself and Djapian in this case plays role only as Django-compatible adaptation.

  • High-level DSL for indexer declaration
  • Result filtering with Django ORM like API
  • Result set compatible with standard Django Paginator
  • Indexing of field, method results and related model attributes
  • Entry filtering before indexing (by trigger function)
  • Results filtering with boolean lookups support
  • Term tagging
  • Spelling corrections
  • Stemming
  • Result ordering by fields
  • Indexers auto discovery
  • Index shell
  • Model changes auto tracking
  • Support for different index spaces

###Usage example

Assume that we have this models in our imaginary application:

class Person(models.Model):
    name = models.CharField(max_length=150)

    def __unicode__(self):
        return self.name

class Entry(models.Model):
    author = models.ForeignKey(Person, related_name="entries")
    title = models.CharField(max_length=250)
    created_on = models.DateTimeField(default=datetime.now)

    is_active = models.BooleanField(default=True)

    text = models.TextField()

    editors = models.ManyToManyField(Person, related_name="edited_entries")

    def headline(self):
        return "%s - %s" % (self.author, self.title)

    def __unicode__(self):
        return self.title

And we want to apply indexing functionality for model Entry. The next step is to create Indexer instance with proper settings. Indexer may look like this:

import djapian

class EntryIndexer(djapian.Indexer):
    fields=["text"]
    tags=[
        ("author",  "author.name" ),
        ("title",   "title",     3),
        ("date",    "created_on"  ),
        ("active",  "is_active"   ),
        ("editors", "editors"     )
    ]
    trigger=lambda indexer, obj: obj.is_active

djapian.space.add_index(Entry, EntryIndexer, attach_as="indexer")

In the django shell create some instances of models:

>>> p = Person.objects.create(name="Alex")
>>> Entry.objects.create(author=p, title="Test entry", text="Not large text field")
>>> Entry.objects.create(author=p, title="Another test entry", is_active=False)
>>> Entry.objects.create(author=p, title="Third small entry", text="Some another text")

>>> Entry.indexer.update()

Thats all! Each Entry instance has been indexed and now ready for search. Let's try:

>>> result = Entry.indexer.search('title:entry')
>>> len(result), result.count()
2, 2
>>> for row in result:
...   row.percent, row.instance.headline()
... 
99 Alex - Test entry
98 Alex - Third small entry

You can follow complete Tutorial for study Djapian basics.