Skip to main content

Pilot is a python library for traversing object trees and graphs

Project description


Pilot is a python library for traversing object trees and graphs. It is currently in pre-alpha.

  1. Usage

  2. Documentation


Download and include the library:

$ pip install pilot

The first step is to define a callback:

>>> def my_callback(node):
...    print node.key, "-->", node.val

Now, define the pilot object and execute a flight:

>>> data = {'test': 1}
>>> pilot = Pilot(my_callback)
>>> new_data =
None --> {'test': 1}
test --> 1
>>> type(new_data)
<class 'pilot.node.dict_extended__node'>
>>> new_data.__node__
<pilot.node.Node object at 0x026CCE90>

See the documentation for more details!


Creating a pilot: pilot = Pilot(*callbacks, **settings)

Callbacks and keyword arguments (settings) can be passed into the pilot constructor.

Settings options:

  • run_callbacks (true|false): Set this to false to skip callbacks completely.

  • callbacks: an array of detailed callback objects. See the Callback section for more information. If this key is specified and there are callbacks passed into the constructor via args, an exception will be raised. If finer control on callbacks is required, use this setting.

  • traversal_mode: the mode for traversing the tree. Options are depth for depth-first processing and breadth for breadth-first processing.

  • structure: Options are tree (default) and graph. Graphs add nodes as “neighbors” rather than parent/child.

  • node_visit_limit: An integer that defines how many times to allow visits to nodes. (Only applicable to graphs.) Set to -1 to allow infinite. Note that infinite trees will process indefinitely - use Pilot.halt() in a callback to end the processing manually.

  • callback_sort: A function that will sort callbacks into an execution order. Note that callback sorting will only apply to a set of callbacks running on a property and position basis, since that is the callback execution stack grouping.

  • callback_sort_mode (0|1): This parameter describes the callback sorting function’s required args. Options are 0 for a standard lambda sort function and 1 for a cmp function.

The configuration defaults to the following:

structure = "Tree"
node_visit_limit = 1
traversal_mode = 'depth'
run_callbacks = True
callback_sort: lambda self, x: x.priority
callback_sort_mode = 0, object, rootkey=None, rootpath=None, rootparent=None)

Traverse an object and execute callbacks during the traversal. The flight will convert every property into a Node, and will make that node available on the property through property.__node__. Note that this means the process will ocassionally change the underlying classes of objects to support adding the __node__ property. As of now, most types are supported – if the conversion process fails, it may be because Pilot is unable to modify the underlying class. See the documentation on the Node class for more details.

Pilot.halt() (static method):

Calling this method within a callback will halt processing completely. This allows for early exit of the traversal, which is required for limited processing of infinite trees.


Callbacks are special functions that execute on the object tree/graph as it is being traversed, and are constructed from predefined functions. Each callback executes at different times based on several configuration options listed below. Callbacks are passed a single argument, the Node for the object. The general form of a callback object is:

def my_callback(node):

from pilot import Callback
callback = Callback(my_callback)

See the “Usage” section of the documentation for a shorthand way to apss callbacks into the Pilot.

Here’s an example, using custom callbacks:

data = [
    {'name': 'Jim', 'friends': [{},{}]},
    {'name': 'Jane', 'friends': [{},{},{}]},
    {'name': 'Joe', 'friends': []}

def count_friends(node):
    except: # not a person
    person = node.val
    friends = person.get('friends', None)
    if friends:
        print person.get("name") + " has " + str(len(friends)) + " friends."
        print person.get("name") + " has no friends."

from pilot import Callback, Pilot
callback = Callback(count_friends, containers=['dict'])
pilot = Pilot(callbacks=[callback])
newdata =

Here are the keyword args you can supply in a callback configuration, most of which act as filters:

  • containers: a list of containers to run on. Options are 'dict', 'list', and 'value'. If unspecifed, the callback will run on any container.

  • keys: an array of keys to run on. The callback will check the key of the property against this list. If unspecified, the callback will run on any key.

  • positions: a list of positions in the traversal to run on. Options are 'pre' (before any list/object is traversed), and 'post' (after any list/object is traversed). For properties of container-type 'value', these two run in immediate succession. If unspecifed, the callback will run 'post'.

  • priority: an integer that represents the callback’s place in the execution order. Lower priorities execute first.

  • run_for_orphans: a boolean that determines whether the callback should run on oprhan nodes.

If any of the filters are not sufficient, additional checks can be done in the callback.


Node objects represent a single node in the tree, providing metadata about the value, its parents, siblings, children, and more. Nodes have the following properties:

  • key: The key of this property as defined on it’s parent. For example, if this callback is running on the 'weight' property of a person, the key would be 'weight'. Note that this will be None for properties in list.

  • value: The value of the property. To use the above example, the value would be something like '183'.

  • container: The type of container of the property expressed in an integer. See the following enumeration for container types: from pilot.definitions import ContainerType

  • encountered: The number of times the pilot has processed the node.

  • id: A unique id for the node (within the context of piot).

You may also call the following methods off of a node object. Keep in mind that some of the node accessors may not be populated based on where you are in your traversal. If you want to wait for the relationship tree to be fully established, run operations on the output of a flight, which will be a data object with each value annotated with a __node__ property:

  • is_orphan(): Returns a boolean of whether the node is an orphan (no parents.)

Node relationship accessors

  • parent(): The first node under which the property exists. node.parent is another instance of node, and will have all the same properties. This is a convenience method that’s useful in trees, where only one parent is possible.

  • parents(): A method that returns parents of a node. An optional search parameters object can be passed in to filter the list to all who have matching key-values. For example, node.parents(key='name', val='Tom') will return all parents where key == 'name' and val == 'Tom'.

  • children(): A method that returns children of a node (i.e. all nodes whose parent is this node.) An optional search parameters object can be passed in to filter the list to all who have matching key-values. For example, node.children(key='name', val='Tom') will return all children where key == 'name' and val == 'Tom'.

  • neighbors(): A method that returns adjacent (non-parent, non-child) nodes, for use in graphs. An optional search parameters object can be passed in to filter the list to all who have matching key-values. For example, node.neighbors(key='name', val='Tom') will return all neighbors where key == 'name' and val == 'Tom'.

  • siblings(): A method that returns all nodes that exist alongside the current node within its parents. For parents of container 'object', this includes all other properties of the parent object. For parents of type 'array', this includes all other nodes in that array.

  • orphans(): A method that returns all connected root nodes.

  • ancestors(): A method that returns a list of all ancestor nodes, going back to the root.

  • descendants(): A method that returns a list of all descendant nodes.

Each relationship accessor can be pased the following keyword arguments:

  • filters=None: A dictionary of key:val that will be matched against each encountered node’s val (as of now this is only useful for dict types).

  • as_value=False: Boolean that will make the call return node values rather than node objects.

  • as_generator=False: Boolean that will make the call return a generator object for the relationship. This is useful when the relationship list is large or inifinite, since it will not attempt to build the list entirely at call time.

Project details

Download files

Download the file for your platform. If you're not sure which to choose, learn more about installing packages.

Source Distributions

No source distribution files available for this release.See tutorial on generating distribution archives.

Built Distribution

pilot-0.0.6-py2-none-any.whl (15.7 kB view hashes)

Uploaded Python 2

Supported by

AWS AWS Cloud computing and Security Sponsor Datadog Datadog Monitoring Fastly Fastly CDN Google Google Download Analytics Microsoft Microsoft PSF Sponsor Pingdom Pingdom Monitoring Sentry Sentry Error logging StatusPage StatusPage Status page