Skip to main content

Easily traverse nested Python data structures

Project description

roam – Easily traverse nested Python data structures

roam provides an API to more easily traverse nested data structures using standard Python syntax without pesky error-handling at each step.

Build Status black

There are three simple steps to use roam:

  1. Wrap your data in a Roamer shim
  2. Express the path or paths to traverse through your data with "dot" or "slice" notation, whichever you prefer
  3. Get the result by calling the Roamer shim object like a function.
# Example nested data: nested dicts and class with attributes
>>> import collections
>>> Point = collections.namedtuple('Point', ['x', 'y'])
>>> data = {"a": {"b": {"c": Point(100, 200)}}}

# 1. Wrap your data in a Roamer shim
>>> import roam
>>> roamer = roam.Roamer(data)

# 2. Express path to traverse
>>> step = roamer.a.b.c.x

# 3. Get result by calling the Roamer shim
>>> step()

# Put it all together slightly differently (read on for details)
>>> roam.r(data).a.b.c['y']()


Install roam with pip:

$ pip install roam

roam works with Python versions 3.6 and later and has no dependencies.


Roamer shim

roam works by providing the Roamer class as a shim over your data objects, to intercepts Python operations and do some extra work to make it easier to traverse nested data.

Get a shim object over your data by calling roam.Roamer(data) or you can use the shorter r alias: roam.r(data)

Traverse paths

You traverse your data within the roam shim by expressing the path (or paths) to follow in Python attribute (dot) or key/index (slice) syntax.

At each step you express in a path, roam returns a new Roamer shim that represents data at that point in the path and the steps taken up to there.

Because roam intercepts and interprets the path operations it can provide some nice features:

  • use dot syntax whether the data item supports attribute or index lookups:

    >>> roam.r({"key": "value"}).key()
  • use slice syntax if you prefer, roam makes dot or slice operations work regardless of the underlying objects:

    >>> roam.r(Point(x=1, y=2))["x"]()
  • mix and match dot and slice to your heart's content:

    >>> roam.r({"point": Point(x=1, y=2)}).point["y"]()
  • use slice syntax to traverse a path step that cannot be a valid Python attribute name:

    >>> roam.r({"no-dash-in-attrs": "thanks"})["no-dash-in-attrs"]()

Generally it makes no difference whether you choose dot or slice syntax to traverse a path, but in cases where an attribute and a key have the same name the choice can matter. Because roam applies your chosen operation first, you can handle this situation by telling it what to do:

# Data with ambiguous "items" name: keyword in dict, and dict method
>>> roamer = roam.r({"items": [1, 2, 3]})

# A dot lookup returns the dict method, which probably isn't what you want...
>>> roamer.items()
dict_items([('items', [1, 2, 3])])

# use a slice lookup instead. Roam will then do a slice lookup first
>>> roamer["items"]()
[1, 2, 3]

Get a result, or MISSING

You get a final result by calling the shim Roamer object like a function with () parentheses, to tell roam to return the underlying data from behind the shim.

If you expressed a valid path through your data you will get the result you expect.

If you expressed an invalid path, roam will not complain or raise an exception. Instead, it will return a roam.MISSING marker object to let you know that there is no data available at the path.

The roam.MISSING object is falsey in a number of ways, so you can either check for an invalid "missing" result directly or rely on its falsey behaviour:

>>> roamer = roam.r(Point(x=1, y=2))

# Check for the `roam.MISSING` object directly
>>> roamer.z() is roam.MISSING

# Check indirectly via falsey behaviour
>>> bool(roamer.z())
>>> len(roamer.z())
>>> [i for i in roamer.z()]

# The falsey MISSING object makes it easy to fall back to a default
>>> roamer.x() or "My fallback"
>>> roamer.z() or "My fallback"
'My fallback'

Of course, sometimes it's better to fail very clearly with an exception. Use the _raise argument to trigger a rich RoamPathException instead of returning a roam.MISSING object:

>>> try:
...     roamer.x.y.z(_raise=True)
... except roam.RoamPathException as ex:
...     str(ex)
'<RoamPathException: missing step 2 .y for path <Point>.x.y.z at <int>>'

Traverse collections

If your data includes collections of items such as a list, you can tell roam to iterate over the collection and apply following path lookups to each item in the collection instead of the collection as a whole.

You do this with a standard slice operation that would return a collection in standard Python usage. Use the special [:] slice to iterate over all items in the collection, or a subset slice using [2:3] etc to iterate over a subset.

When you traverse a collection with a slice operation, the final result is a tuple of data items.

For example:

>>> roamer = roam.r({
...     "people": [
...         {"name": "Alice", "age": 34},
...         {"name": "Bob", "age": 42},
...         {"name": "Trudy"},  # Unknown age
...     ]
... })

# A `list` object does not have the attributee `name`

# Use the "all items" [:] slice operation to iterate over each item
>>> roamer.people[:].name()
('Alice', 'Bob', 'Trudy')

# Get all but the last person names
>>> roamer.people[:-1].name()
('Alice', 'Bob')

roam handles collections differently from single items in the path in that it ignores items where the following path is invalid, filtering them out instead of returning roam.MISSING marker objects.

You can think of a collection traversal in roam as being like a combined for-each and filter.

# Alice is 34, Bob is 42, Trudy has no "age" data
>>> roamer.people[:].age()
(34, 42)

When traversing a collection, if you use an integer index lookup instead of a slice roam will return the single nth item from the collection as you would expect:

>>> roamer.people[-1].name()

WARNING: roam has only rudimentary support for traversing nested collections. Simple cases should work, but if you need to traverse non-trivial collections data you should do the work with for loops in your code.

>>> roamer = roam.r({
...     "people": [
...         {"name": "Alice", "pets": [
...             {"type": "cat", "name": "Mog"},
...             {"type": "dog", "name": "Spot"},
...         ]},
...         {"name": "Bob", "pets": [
...             {"type": "budgie", "name": "Bertie"},
...         ]},
...     ]
... })

# We can get the names of the "pets" collection under the people "collection"
>>> roamer.people[:]
('Mog', 'Spot', 'Bertie')

# And look up just the n-th result at at a given level
>>> roamer.people[:][0]()



  • falsey-ness of roam.MISSING
  • rich path descriptions and exceptions
  • roam is a shim (lookup order, equality, iteration, falsey-ness, truthiness, length)
  • call nested methods
  • re-wrap result of nested method call with the _roam option
  • fail fast with the _raise option
  • call arbitrary functions with the _invoke option

Related projects

These similar tools and libraries helped inspire and inform roam:

  • Django template language's variable dot lookup
  • glom – "Restructuring data, the Python way."
  • traversify – "Handy python classes for manipulating json data, providing syntactic sugar for less verbose, easier to write code."



roam is licensed under Apache, Version 2.0

Copyright 2019 James Murty

Licensed under the Apache License, Version 2.0 (the "License");
you may not use this file except in compliance with the License.
You may obtain a copy of the License at

Unless required by applicable law or agreed to in writing, software
distributed under the License is distributed on an "AS IS" BASIS,
See the License for the specific language governing permissions and
limitations under the License.

Project details

Download files

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

Source Distribution

roam-0.2.tar.gz (21.0 kB view hashes)

Uploaded source

Built Distribution

roam-0.2-py3-none-any.whl (26.9 kB view hashes)

Uploaded py3

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