dict2xml

dict2xml

Super Simple utility to convert a python dictionary into an xml string

Installation

Install using pip::

python -m pip install dict2xml

example

.. code-block:: python

from dict2xml import dict2xml

data = { 'a': 1, 'b': [2, 3], 'c': { 'd': [ {'p': 9}, {'o': 10} ], 'e': 7 } }

print dict2xml(data, wrap="all", indent=" ")

Output

.. code-block:: xml

1 2 3

9

10 7

methods

dict2xml.dict2xml(data, *args, **kwargs) Equivalent to:

.. code-block:: python

    dict2xml.Converter(*args, **kwargs).build(data)

dict2xml.Converter(wrap="", indent=" ", newlines=True) Knows how to convert a dictionary into an xml string

* wrap: Wraps the entire tree in this tag
* indent: Amount to prefix each line for each level of nesting
* newlines: Whether or not to use newlines

dict2xml.Converter.build(data, iterables_repeat_wrap=True, closed_tags_for=None, data_sorter=None) Instance method on Converter that takes in the data and creates the xml string

* iterables_repeat_wrap - when false the key the array is in will be repeated
* closed_tags_for - an array of values that will produce self closing tags
* data_sorter - an object as explained below for sorting keys in maps

dict2xml.DataSorter An object used to determine the sorting of keys for a map of data.

By default an ``OrderedDict`` object will not have it's keys sorted, but any
other type of mapping will.

It can be made so even ``OrderedDict`` will get sorted by passing in
``data_sorter=DataSorter.always()``.

Or it can be made so that keys are produced from the sorting determined by
the mapping with ``data_sorter=DataSorter.never()``.

.. note:: When this library was first created python did not have deterministic
   sorting for normal dictionaries which is why default everything gets sorted but
   ``OrderedDict`` do not.

To create custom sorting logic requires an object that has a single ``keys_from``
method on it that accepts a map of data and returns a list of strings, where only
the keys that appear in the list will go into the output and those keys must exist
in the original mapping.

Self closing tags

To produce self closing tags (like <item/>) then the build method must be given a list of values under closed_tags_for. For example, if you want None to produce a closing tag then:

.. code-block:: python

example = {
    "item1": None,
    "item2": {"string1": "", "string2": None},
    "item3": "special",
}

result = Converter("").build(example, closed_tags_for=[None])
assert result == dedent("""
    <item1/>
    <item2>
        <string1></string1>
        <string2/>
    </item2>
    <item3>special</item3>
""").strip())

Here only string2 gets a self closing tag because it has data of None, which has been designated as special.

If you want to dynamically work out which tags should be self closing then you may provide an object that implements __eq__ and do your logic there.

Limitations

• No attributes on elements
• Currently no explicit way to hook into how to cope with your custom data
• Currently no way to insert an xml declaration line

Changelog

1.7.7 - 10 July 2025 * Converted the tests to plain python to remove the noseOfYeti dependency

1.7.6 - 8 August 2024 * Fixed the dict2xml.dict2xml entry point to distribute options correctly

1.7.5 - 13 February 2024 * Introduced the data_sorter option

1.7.4 - 16 January 2024 * Make the tests compatible with pytest8

1.7.3 - 25 Feb 2023 * This version has no changes to the installed code. * This release converts to hatch for packaging and adds a wheel to the package on pypi. * CI will now run against python 3.11 as well

1.7.2 - 18 Oct 2022 * This version has no changes to the installed code. * This release adds the tests to the source distribution put onto pypi.

1.7.1 - 16 Feb 2022 * Adding an option to have self closing tags when the value for that tag equals certain values

1.7.0 - 16 April, 2020 * Use collections.abc to avoid deprecation warning. Thanks @mangin. * This library no longer supports Python2 and is only supported for Python3.6+. Note that the library should still work in Python3.5 as I have not used f-strings, but the framework I use for the tests is only 3.6+.

1.6.1 - August 27, 2019 * Include readme and LICENSE in the package

1.6 - April 27, 2018 * No code changes * changed the licence to MIT * Added more metadata to pypi * Enabled travis ci * Updated the tests slightly

1.5 * No changelog was kept before this point.

Development

To enter a virtualenv with dict2xml and dev requirements installed run::

> source run.sh activate

Tests may be run with::

> ./test.sh 

Or::

> ./run.sh tox

Linting and formatting is via::

> ./format
> ./lint