Skip to main content

Converts XML into JSON/Python dicts/arrays and vice-versa.

Project description

xmljson converts XML into Python dictionary structures (trees, like in JSON) and vice-versa.


XML can be converted to a data structure (such as JSON) and back. For example:

        <name value="Alice"/>
        <name value="Bob"/>

can be converted into this data structure (which also a valid JSON object):

{ "employees": [
    { "person": {
        "name": {"@value": "Alice"}
    } },
    { "person": {
        "name": {"@value": "Alice"}
    } }
] }

This uses the BadgerFish convention that prefixes attributes with @. The conventions supported by this library are:

  • Abdera: Use "attributes" for attributes, "children" for nodes

  • BadgerFish: Use "$" for text content, @ to prefix attributes

  • Cobra: Use "attributes" for attributes (even when empty), "children" for nodes, values are strings

  • GData: Use "$t" for text content, attributes added as-is

  • Yahoo Use "content" for text content, attributes added as-is

  • Parker: Use tail nodes for text content, ignore attributes

Convert data to XML

To convert from a data structure to XML using the BadgerFish convention:

>>> from xmljson import badgerfish as bf
>>> bf.etree({'p': {'@id': 'main', '$': 'Hello', 'b': 'bold'}})

This returns an array of etree.Element structures. In this case, the result is identical to:

>>> from xml.etree.ElementTree import fromstring
>>> [fromstring('<p id="main">Hello<b>bold</b></p>')]

The result can be inserted into any existing root etree.Element:

>>> from xml.etree.ElementTree import Element, tostring
>>> result = bf.etree({'p': {'@id': 'main'}}, root=Element('root'))
>>> tostring(result)
'<root><p id="main"/></root>'

This includes lxml.html as well:

>>> from lxml.html import Element, tostring
>>> result = bf.etree({'p': {'@id': 'main'}}, root=Element('html'))
>>> tostring(result, doctype='<!DOCTYPE html>')
'<!DOCTYPE html>\n<html><p id="main"></p></html>'

For ease of use, strings are treated as node text. For example, both the following are the same:

>>> bf.etree({'p': {'$': 'paragraph text'}})
>>> bf.etree({'p': 'paragraph text'})

By default, non-string values are converted to strings using Python’s str, except for booleans – which are converted into true and false (lower case). Override this behaviour using xml_fromstring:

>>> tostring(bf.etree({'x': 1.23, 'y': True}, root=Element('root')))
>>> from xmljson import BadgerFish              # import the class
>>> bf_str = BadgerFish(xml_tostring=str)       # convert using str()
>>> tostring(bf_str.etree({'x': 1.23, 'y': True}, root=Element('root')))

Convert XML to data

To convert from XML to a data structure using the BadgerFish convention:

>>>'<p id="main">Hello<b>bold</b></p>'))
{"p": {"$": "Hello", "@id": "main", "b": {"$": "bold"}}}

To convert this to JSON, use:

>>> from json import dumps
>>> dumps('<p id="main">Hello<b>bold</b></p>')))
'{"p": {"b": {"$": "bold"}, "@id": "main", "$": "Hello"}}'

To preserve the order of attributes and children, specify the dict_type as OrderedDict (or any other dictionary-like type) in the constructor:

>>> from collections import OrderedDict
>>> from xmljson import BadgerFish              # import the class
>>> bf = BadgerFish(dict_type=OrderedDict)      # pick dict class

By default, values are parsed into boolean, int or float where possible (except in the Yahoo method). Override this behaviour using xml_fromstring:

>>> dumps('<x>1</x>')))
'{"x": {"$": 1}}'
>>> bf_str = BadgerFish(xml_fromstring=False)   # Keep XML values as strings
>>> dumps('<x>1</x>')))
'{"x": {"$": "1"}}'
>>> bf_str = BadgerFish(xml_fromstring=repr)    # Custom string parser
'{"x": {"$": "\'1\'"}}'

xml_fromstring can be any custom function that takes a string and returns a value. In the example below, only the integer 1 is converted to an integer. Everything else is retained as a float:

>>> def convert_only_int(val):
...     return int(val) if val.isdigit() else val
>>> bf_int = BadgerFish(xml_fromstring=convert_only_int)
>>> dumps('<p><x>1</x><y>2.5</y><z>NaN</z></p>')))
'{"p": {"x": {"$": 1}, "y": {"$": "2.5"}, "z": {"$": "NaN"}}}'


To use a different conversion method, replace BadgerFish with one of the other classes. Currently, these are supported:

>>> from xmljson import abdera          # == xmljson.Abdera()
>>> from xmljson import badgerfish      # == xmljson.BadgerFish()
>>> from xmljson import cobra           # == xmljson.Cobra()
>>> from xmljson import gdata           # == xmljson.GData()
>>> from xmljson import parker          # == xmljson.Parker()
>>> from xmljson import yahoo           # == xmljson.Yahoo()


Conventions may support additional options.

The Parker convention absorbs the root element by default. preserves the root instance:

>>> from xmljson import parker, Parker
>>> from xml.etree.ElementTree import fromstring
>>> from json import dumps
>>> dumps('<x><a>1</a><b>2</b></x>')))
'{"a": 1, "b": 2}'
>>> dumps('<x><a>1</a><b>2</b></x>'), preserve_root=True))
'{"x": {"a": 1, "b": 2}}'


This is a pure-Python package built for Python 2.6+ and Python 3.0+. To set up:

pip install xmljson


  • Test cases for Unicode

  • Support for namespaces and namespace prefixes


0.1.8 (9 May 2017)

  • Add Abdera and Cobra conventions

  • Add option to preserve root element in Parker convention.

Thanks to @dagwieers

0.1.6 (18 Feb 2016)

  • Add xml_fromstring= and xml_tostring= parameters to constructor to customise string conversion from and to XML.

0.1.5 (23 Sep 2015)

  • Add the Yahoo XML to JSON conversion method.

0.1.4 (20 Sep 2015)

  • Fix GData.etree() conversion of attributes. (They were ignored. They should be added as-is.)

0.1.3 (20 Sep 2015)

  • Simplify {'p': {'$': 'text'}} to {'p': 'text'} in BadgerFish and GData conventions.

  • Add test cases for .etree() – mainly from the MDN JXON article.

  • dict_type/list_type do not need to inherit from dict/list

0.1.2 (18 Sep 2015)

  • Always use the dict_type class to create dictionaries (which defaults to OrderedDict to preserve order of keys)

  • Update documentation, test cases

  • Remove support for Python 2.6 (since we need collections.Counter)

  • Make the Travis CI build pass

0.1.1 (18 Sep 2015)

  • Convert true, false and numeric values from strings to Python types

  • is compliant with Parker convention (bugs resolved)

0.1.0 (15 Sep 2015)

  • Two-way conversions via BadgerFish, GData and Parker conventions.

  • First release on PyPI.

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

xmljson-0.1.8.tar.gz (23.5 kB view hashes)

Uploaded source

Built Distribution

xmljson-0.1.8-py2.py3-none-any.whl (11.6 kB view hashes)

Uploaded py2 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