PostgreSQL LTreeField for Django

Minimalist Django Field for the PostgreSQL ltree Type.

django-ltree-field attempts to make very few assumptions about your use case.

For a higher level API based on django-ltree-field, consider using a prebuilt model from django-ltree-utils.

It should be possible to re-implement the django-treebeard API, allowing for drop-in compatibility, but that is not a specific goal at this time. If someone starts this, let me know and I will provide some assistance.

Documentation

The full documentation is at https://django-ltree-field.readthedocs.io.

Quickstart

Install PostgreSQL LTreeField for Django:

pip install django-ltree-field

Add it to your `INSTALLED_APPS`:

INSTALLED_APPS = (
    ...
    'django_ltree_field',
    ...
)

Add an LTreeField to a new or existing model:

from django_ltree_field.fields import LTreeField

class SimpleNode(models.Model):
    path = LTreeField(index=True, unique=True)

    class Meta:
        ordering = ['path']

Features

• Implements only the bare minimum to make the ltree PostgreSQL type usable
• LTreeField accepts a string of dotted labels, or a list of labels
• The ltree type is adapted to a Python list
• Relatively complete set of lookups and transforms.

Non-Features

• Does not implement an abstract "Node" model which has a nicer API (See django-ltree-utils for ready-made classes and managers)
• Does virtually no sanity checking. You can insert nodes without roots, and generally put the tree in a bad state
• PostgreSQL compatibility only

Future Features

I will happily accept minimal features required to make the field be reasonably usable. In particular, every operator, function, and example on the official PostgreSQL docs should be implemented with Django's ORM, with no RawSQL or non-idiomatic code.

Higher-level or richer features should be contributed to django-ltree-utils. As a rule of thumb, if an operation requires referencing more than one row at a time, or maintaining some more complicated state, it probably belongs there.

Running Tests

You need to have a reasonably updated version of PostgreSQL listening on port 5444. You can use docker-compose to start a server

docker-compose up

Does the code actually work?

source <YOURVIRTUALENV>/bin/activate
(myenv) $ pip install -r requirements.txt -r requirements_test.txt --upgrade
(myenv) $ ./runtests.py