PostgreSQL Ltree trees (experimental)

This is an efficient tree implementation using PostgreSQL’s ltree module. It requires a PostgreSQL database.

This is currently an experimental implementation, open for testing and feedback from the community.

Treebeard uses a simple alphabet to generate path hierarchies for objects in the database. In order to ensure efficient ordering, it uses an approach similar to the Materialized Path implementation to use path values that match the desired order of nodes in the database.

To use the ltree module, you need to create the extension in your database:

CREATE EXTENSION IF NOT EXISTS ltree;

Warning

As with all tree implementations, please be aware of the Known Caveats.

Inheritance diagram of LT_Node
class treebeard.ltree.LT_Node(*args, **kwargs)

Bases: Node

Abstract model to create your own Postgres LTree trees.

Warning

Do not change the values of path directly. Use one of the included methods instead. Consider these values read-only.

Warning

Do not change the values of node_order_by after saving your first object. Doing so will result in objects being ordered incorrectly.

Warning

If you need to define your own Manager class, you’ll need to subclass LT_NodeManager.

Also, if in your manager you need to change the default queryset handler, you’ll need to subclass LT_NodeQuerySet.

Example:

class SortedNode(LT_Node):
   node_order_by = ['numval', 'strval']

   numval = models.IntegerField()
   strval = models.CharField(max_length=255)

Read the API reference of treebeard.models.Node for info on methods available in this class, or read the following section for methods with particular arguments or exceptions.

node_order_by

Attribute: a list of model fields that will be used for node ordering. When enabled, all tree operations will assume this ordering. This takes precedence over drag and drop ordering in the Django admin.

Example:

node_order_by = ['field1', 'field2', 'field3']

Warning

node_order_by values are used to determine correct node ordering before

an object is inserted/moved. This means any fields that are auto-populated at a database level, e.g., AutoField(), or DateTimeField(auto_now=True) will be ignored for the purpose of ordering if a value isn’t provided manually.

path

ltree field, stores an ltree hierarchy for the node. The values are auto-generated by Treebeard from a simple alphabet.

classmethod add_root(**kwargs)

Adds a root node to the tree.

This method saves the node in database. The object is populated as if via:

` obj = cls(**kwargs) `

See: treebeard.models.Node.add_root()

add_child(**kwargs)

Adds a child to the node.

This method saves the node in database. The object is populated as if via:

` obj = self.__class__(**kwargs) `

See: treebeard.models.Node.add_child()

add_sibling(pos=None, **kwargs)

Adds a new node as a sibling to the current node object.

This method saves the node in database. The object is populated as if via:

` obj = self.__class__(**kwargs) `

See: treebeard.models.Node.add_sibling()

move(target, pos=None)

Moves the current node and all it’s descendants to a new position relative to another node.

See: treebeard.models.Node.move()

classmethod get_tree(parent=None)
Returns:

A queryset of nodes ordered as DFS, including the parent. If no parent is given, the entire tree is returned.

See: treebeard.models.Node.get_tree()

class treebeard.ltree.LT_NodeManager(*args, **kwargs)

Bases: Manager

Custom manager for nodes in a Materialized Path tree.

class treebeard.ltree.LT_NodeQuerySet(model=None, query=None, using=None, hints=None)

Bases: QuerySet

Custom queryset for the tree node manager.

Needed only for the custom delete method.