d1ct 1.0.3rc8

dict implementation supporting attribute access, recursion, json, hashable, subclass, metaclass

d1ct - a hybrid object/mapping

FEATURES

• directly inherited from dict
• accessible by attribute. o.x == o['x']
• This works also for all corner cases.
• json.dumps json.loads
• vars() works on it!
• hashable! (should not yet trust on it, though)
• autocomplete works even if the objects comes from a list
• best of all, it works recursively. So all inner dicts will become d1cts too.

ehhh, no downsides?

oh hell yeah it got downsides

the performance which is a magnitude slower (~6x) than builtin dict
to give you an impression of timeit results:

d1ct

0.26289230000111274
0.2543108999961987
0.25743720002355985
0.2560539000260178
0.2556736999831628
0.25478869999642484
0.2541010999993887
0.2552008999919053
0.2651131000020541
0.25846979999914765

dict

0.04928750000544824
0.04902900001616217
0.049049399996874854
0.050644999981159344
0.0496991999971214
0.04940899999928661
0.0497540999785997
0.04959690000396222
0.04939249999006279
0.04959330000565387

So... this implementation might have side effects which I did not notice, because of different use cases or whatever.

Usage / Testing

import d1ct

d = d1ct({"a": 1, "b": {"b1": "c1"}}, kw1=1, kw2=2)
d["z1"] = {}
d.z1.update({"yyyy": 2323232})

assert d.a == d['a'] == 1
assert d.z1 == d['z1'] == {'yyyy': 2323232}
assert d == {'a': 1, 'b': {'b1': 'c1'}, 'kw1': 1, 'kw2': 2, 'z1': {'yyyy': 2323232}}
assert type(d.z1) is d1ct

d.z1.update({"yyyy": 2323232})
assert type(d.z1.yyyy) is int

d.z1.update({"yyyy": {"xbb": [1, 2, 23]}})
assert type(d.z1.yyyy) is d1ct

import json
json.dumps(d)

d |= {'more_keys': {'really': 'yeah!'}}

assert d == {'a': 1,'b': {'b1': 'c1'}, 'kw1': 1,'kw2': 2, 'z1': {'yyyy': {'xbb': [1, 2, 23] } },'more_keys': {'really': 'yeah!'} }