How do I serialize a Python dictionary into a string, and then back to a dictionary?

Question

How do I serialize a Python dictionary into a string, and then back to a dictionary? The dictionary will have lists and other dictionaries inside it.

Kalin

Are you familiar with pickle? – Indian

a module that is part of the Python Standard Library – Armindaarming

Answer 1

It depends on what you're wanting to use it for. If you're just trying to save it, you should use pickle (or, if you’re using CPython 2.x, cPickle, which is faster).

>>> import pickle
>>> pickle.dumps({'foo': 'bar'})
b'\x80\x03}q\x00X\x03\x00\x00\x00fooq\x01X\x03\x00\x00\x00barq\x02s.'
>>> pickle.loads(_)
{'foo': 'bar'}

If you want it to be readable, you could use json:

>>> import json
>>> json.dumps({'foo': 'bar'})
'{"foo": "bar"}'
>>> json.loads(_)
{'foo': 'bar'}

json is, however, very limited in what it will support, while pickle can be used for arbitrary objects (if it doesn't work automatically, the class can define __getstate__ to specify precisely how it should be pickled).

>>> pickle.dumps(object())
b'\x80\x03cbuiltins\nobject\nq\x00)\x81q\x01.'
>>> json.dumps(object())
Traceback (most recent call last):
  ...
TypeError: <object object at 0x7fa0348230c0> is not JSON serializable

Jdavie

I wish I knew what on earth that -1 was for. – Jdavie

I guess this -1 might be for not mentioning security problems inherent in pickling. See https://mcmap.net/q/243677/-attacking-python-39-s-pickle – Judon

It is worth mentioning that the cPickle part of the answer is not relevant for python 3.x. See here for the official explanation. In short, the accelerated C version of a package should be the default choice for any python module, and, if not available, the module itself falls back to the python implementation. This encapsulates the implementation from the user. Quote: In Python 3.0... Users should always import the standard version, which attempts to import the accelerated version and falls back to the pure Python version. – Singularity

"Warning The pickle module is not secure. Only unpickle data you trust." - docs – Parra

Answer 2

Pickle is great but I think it's worth mentioning literal_eval from the ast module for an even lighter weight solution if you're only serializing basic python types. It's basically a "safe" version of the notorious eval function that only allows evaluation of basic python types as opposed to any valid python code.

Example:

>>> d = {}
>>> d[0] = range(10)
>>> d['1'] = {}
>>> d['1'][0] = range(10)
>>> d['1'][1] = 'hello'
>>> data_string = str(d)
>>> print data_string
{0: [0, 1, 2, 3, 4, 5, 6, 7, 8, 9], '1': {0: [0, 1, 2, 3, 4, 5, 6, 7, 8, 9], 1: 'hello'}}

>>> from ast import literal_eval
>>> d == literal_eval(data_string)
True

One benefit is that the serialized data is just python code, so it's very human friendly. Compare it to what you would get with pickle.dumps:

>>> import pickle
>>> print pickle.dumps(d)
(dp0
I0
(lp1
I0
aI1
aI2
aI3
aI4
aI5
aI6
aI7
aI8
aI9
asS'1'
p2
(dp3
I0
(lp4
I0
aI1
aI2
aI3
aI4
aI5
aI6
aI7
aI8
aI9
asI1
S'hello'
p5
ss.

The downside is that as soon as the the data includes a type that is not supported by literal_ast you'll have to transition to something else like pickling.

Dy

literal_ast seem to me unable to encode instances of user defined classes, even the simplest one; the solution mentioned by @Drobman (namely pyYAML) do that and produces human readable serializations. From pypi.org project home (June 2021): YAML is a data serialization format designed for human readability and interaction with scripting languages. PyYAML is a YAML parser and emitter for Python. – Auberge

Answer 3

Use Python's json module, or simplejson if you don't have python 2.6 or higher.

Urson

+1: json is way better than pickle and can be used in the same way: json.dumps(mydict) and json.loads(mystring) – Chloechloette

but json can only do strings, numbers, lists, and dictionaries while pickle can do any python type but json is far more portable then pickle for the types it can do – Urson

When you use json.dumps(), take care of some types (False, True, and None) because they are not compatible with json – Omarr

Answer 4

If you fully trust the string and don't care about python injection attacks then this is very simple solution:

d = { 'method' : "eval", 'safe' : False, 'guarantees' : None }
s = str(d)
d2 = eval(s)
for k in d2:
    print k+"="+d2[k]

If you're more safety conscious then ast.literal_eval is a better bet.

Sansen

honestly this is the method I use all the time. thanks for sharing the safety tip. I use repr instead of str if the dictionary contains custom made objects that can be initialized by the repr string – Everyday

You should use ast.literal_eval by default. eval has zero added values and a big security issue. – Toscanini

Bad things happen because persons honestly thought there is no security concerns in their particular peace of code, so they can just happly eval away. I'm just disgusted every time, someone promote this culture of sloppiness. Just use json.dumps and json.loads (or any other non-eval solution), there is no real reason not to – Parra

Answer 5

One thing json cannot do is dict indexed with numerals. The following snippet

import json
dictionary = dict({0:0, 1:5, 2:10})
serialized = json.dumps(dictionary)
unpacked   = json.loads(serialized)
print(unpacked[0])

will throw

KeyError: 0

Because keys are converted to strings. cPickle preserves the numeric type and the unpacked dict can be used right away.

Gwyn

Answer 6

pyyaml should also be mentioned here. It is both human readable and can serialize any python object.
pyyaml is hosted here:
https://pypi.org/project/PyYAML

Drobman

See pypi.org/project/PyYAML. Consider vulnerabilities though (eg: ibm.com/support/pages/…) – Auberge

thx for the new link, I updated my answer accordingly. – Drobman

Answer 7

While not strictly serialization, json may be reasonable approach here. That will handled nested dicts and lists, and data as long as your data is "simple": strings, and basic numeric types.

Sharp

Answer 8

A new alternative to JSON or YaML is NestedText. It supports strings that are nested in lists and dictionaries to any depth. It conveys nesting through the use of indenting, and so has no need for either quoting or escaping. As such, the result tends to be very readable. The result looks like YaML, but without all the special cases. It is especially appropriate for serializing code snippets. For example, here is an a single test case extracted from a much larger set that was serialized with NestedText:

base tests:
    -
        args: --quiet --config test7 files -N configs/subdir
        expected:
            > Archive: test7-\d\d\d\d-\d\d-\d\dT\d\d:\d\d:\d\d
            > «TESTS»/configs/subdir/
            > «TESTS»/configs/subdir/file

Be aware, that integers, floats, and bools are converted to strings.

Seaway

Answer 9

If you are trying to only serialize then pprint may also be a good option. It requires the object to be serialized and a file stream.

Here's some code:

from pprint import pprint
my_dict = {1:'a',2:'b'}
with open('test_results.txt','wb') as f:
    pprint(my_dict,f)

I am not sure if we can deserialize easily. I was using json to serialize and deserialze earlier which works correctly in most cases.

f.write(json.dumps(my_dict, sort_keys = True, indent = 2, ensure_ascii=True))

However, in one particular case, there were some errors writing non-unicode data to json.

Enumeration