Skip to content

Addict dictionary enhanced with ability to track changes -- currently only supports field additions

License

Notifications You must be signed in to change notification settings

ofjustpy/addict-tracking-changes

Repository files navigation

Python versions Tests Status Coverage Status Flake8 Status

Documentation PyPI Downloads Downloads per week GitHub stars

Introduction

HEADS UP Before using the library carefully read the Known bugs and Caveats section below.

Originally, this repository was a fork of addict by Mats Julian Olsen. Overtime, it has substatially diverged in functionality and codebase that it made sense to breakout as its own repository.

The original addict: provides an alternative and succient interface to manipulate a dictionary. This is especially useful when dealing with heavily nested dictionaries. As example (taken from https://github.com/mewwts/addict) a dictionary created using standard python dictionary interface looks as follows:

body = {
    'query': {
        'filtered': {
            'query': {
                'match': {'description': 'addictive'}
            },
            'filter': {
                'term': {'created_by': 'Mats'}
            }
        }
    }
}`

can be summarized to following three lines:

body = Dict()
body.query.filtered.query.match.description = 'addictive'
body.query.filtered.filter.term.created_by = 'Mats'

This repo builds on original addict and adds contraptions to track key additions in the dictionary. This features comes in quite handy in building reactive webapps where one has to respond to all the changes made on the browser. Addict-tracking-changes is the underpinning data-structure in ofjustpy: a python based webframework build from justpy

The functions relevant to tracking changed history are: get_changed_history and clear_changed_history. The get_changed_history returns an iterator of XPath style paths like /a/b/c/e (see Demo example).

Usage and examples

Installation

This project is not on PyPI. Its a simple package with no third party dependency. Simply clone from github and include the source directory in your PYTHONPATH.

Demo example

from addict import Dict
body = Dict()
body.query.filtered.query.match.description = 'addictive'
body.query.filtered.filter.term.created_by = 'Mats'

for changed_path in body.get_changed_history():
    #<work with changed_path>

body.clear_changed_history()

Behaviour when values are instances of container types

addict works as expected when the values of keys are simple data types (such as string, int, float, etc.). However, for container types such as dict, list, tuples, etc. the behaviour is somewhat differs.

  • dicts are treated as opaque object just like simple data types
from addict import Dict

mydict = Dict()
mydict.a.b.c = {'kk': 1}
mydict.a.b.e = {'dd': 1}

for _ in mydict.get_changed_history():
    print(_) 

will print

/a/b/c
/a/b/e

and not

/a/b/cc/kk
/a/b/e/dd
  • lists are seen as container, i.e., get_changed_history will report path for each element of the list
from addict import Dict

mydict = Dict()

mydict.a.b = [1, [1]]
mydict.a.c = [2, [2, [3]]]

get_changed_history will report following paths:

/a/b/0,
/a/b/1/0,
/a/c/0,
/a/c/1/0,
/a/c/1/1/"
  • tuple, namedtuple, sets tuple behave same as dict and are treated as opaque data structure

Known bugs and Caveats

  1. Only tracks field additions. Deletions and updates are not tracked.
  2. freeze doesn't guards against deletions
  3. building dict from another dict as shown in following expression wont' work
cjs_cfg = Dict(other_dict, track_changes=True)

instead use

cjs_cfg = Dict(track_changes = True)
with open("other_dict.pickle", "rb") as fh:
    x = pickle.load(fh)
for _ in oj.dictWalker(x):
    oj.dnew(cjs_cfg, _[0], _[1])
  1. Use TrackedList for tracking nested list
from addict_tracking_changes import Dict, TrackedList


trackerprop = Dict(track_changes = True)
trackerprop.a.b = [1, TrackedList([1], track_changes=True)]
trackerprop.a.c = [2, TrackedList([2, TrackedList([3], track_changes=True)
                                   ], track_changes=True)
                   ]

which when asked changed history will output as follows:

print(list(trackerprop.get_changed_history()))

output

['/a/b/0', '/a/b/1/0', '/a/c/0', '/a/c/1/0', '/a/c/1/1/0']

APIs

API Description
Dict(*args, **kwargs) Initializes a new Dict object
to_dict(self) Converts the Dict object to a regular dictionary
freeze(self, shouldFreeze=True) Freezes the Dict object, making it immutable
unfreeze(self) Unfreezes the Dict object, making it mutable again
get_changed_history(self, prefix="", path_guards=None) Returns an iterator with the changed history of keys
clear_changed_history(self) Clears the changed history of the Dict object
set_tracker(self, track_changes=False) Sets or resets the change tracker for the Dict object

EndNotes

About

Addict dictionary enhanced with ability to track changes -- currently only supports field additions

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published