Converts a Python dictionary or other native data type into a valid XML string.

Supports item (int, float, long, decimal.Decimal, bool, str, unicode, datetime, none and other number-like objects) and collection (list, set, tuple and dict, as well as iterable and dict-like objects) data types, with arbitrary nesting for the collections. Items with a datetime type are converted to ISO format strings. Items with a None type become empty XML elements.

The root object passed into the dicttoxml method can be any of the supported data types.

To satisfy XML syntax, the method prepends an <?xml version="1.0" encoding="UTF-8" ?> element and wraps the output in a <root> ... </root> element. However, this can be disabled to create XML snippets. Alternately, a custom root element can be specified by passing in the optional custom_root=foobar argument.

For lists of items, if each item is also a collection data type (lists, dict), the elements of that item are wrapped in a generic <item> ... </item> element.

Each element includes an optional type attribute with the data type. By default, the type attribute it included but it can be excluded by passing an optional attr_type=False argument when calling the dicttoxml method.

Note: datetime data types are converted into ISO format strings, and unicode and datetime data types get a str attribute.

Python -> XML
integer   int
long      long
float     float
Decimal   number
string    str
unicode   str
datetime  str
None      null
boolean   bool
list      list
set       list
tuple     list
dict      dict

Elements with an unsupported data type raise a TypeError exception.

If an element name is invalid XML, it is rendered with the name "key" and the invalid name is included as a name attribute. E.g. { "^.{0,256}$": "foo" } would be rendered <key name="^.{0,256}$">foo</key>. An exception is element names with spaces, which are converted to underscores.

This module should work in Python 3.6+. You might be able to get it to work in Python 2.x but that is no longer supported.

The dicttoxml module is published on the Python Package Index, so you can install it using pip.

pip install dicttoxml

Alternately, you can download the tarballed installer - dicttoxml-[VERSION].tar.gz - for this package from the dist directory on github and uncompress it. Then, from a terminal or command window, navigate into the unzipped folder and type the command:

python setup.py install

That should be all you need to do.

Once installed, import the library into your script and convert a dict into xml by running the dicttoxml function:

>>> import dicttoxml
>>> xml = dicttoxml.dicttoxml(some_dict)

Alternately, you can import the dicttoxml() function from the library.

>>> from dicttoxml import dicttoxml
>>> xml = dicttoxml(some_dict)

That's it!

Let's say you want to fetch a JSON object from a URL and convert it into XML. Here's how you can do that:

>>> import json
>>> import urllib
>>> import dicttoxml
>>> page = urllib.urlopen('http://quandyfactory.com/api/example')
>>> content = page.read()
>>> obj = json.loads(content)
>>> print(obj)
{u'mylist': [u'foo', u'bar', u'baz'], u'mydict': {u'foo': u'bar', u'baz': 1}, u'ok': True}
>>> xml = dicttoxml.dicttoxml(obj)
>>> print(xml)
<?xml version="1.0" encoding="UTF-8" ?><root><mylist><item type="str">foo</item><item type="str">bar</item><item type="str">baz</item></mylist><mydict><foo type="str">bar</foo><baz type="int">1</baz></mydict><ok type="bool">true</ok></root>

It's that simple.

Instead of creating a full XML document, you can create an XML snippet for inclusion into another XML document.

Continuing with the example from above:

>>> xml_snippet = dicttoxml.dicttoxml(obj, root=False)
>>> print(xml_snippet)
<mylist><item type="str">foo</item><item type="str">bar</item><item type="str">baz</item></mylist><mydict><foo type="str">bar</foo><baz type="int">1</baz></mydict><ok type="bool">true</ok>

With the optional root argument set to False, the method converts the dict into XML without including an <?xml> prolog or a <root> element to enclose all the other elements.

By default, dicttoxml wraps all the elements in a <root> ... </root> element. Starting in version 1.5, you can change the name of the root element to something else by passing an optional custom_root=some_custom_root argument to the dicttoxml method.

Using our example:

>>> xml = dicttoxml.dicttoxml(obj, custom_root='some_custom_root')
>>> print(xml)
<?xml version="1.0" encoding="UTF-8" ?><some_custom_root><mydict><foo>bar</foo><baz>1</baz></mydict><mylist><item>foo</item><item>bar</item><item>baz</item></mylist><ok>true</ok></some_custom_root>

As you can see, the name of the root element has changed to some_custom_root.

Perhaps you want your outputted XML to have a root but no XML declaration. Starting with version 1.7.15, can call dicttoxml with the optional xml_declaration argument set to False:

>>> xml = dicttoxml.dicttoxml(xml_declaration=False)
>>> print(xml)
<root><ok type="bool">true</ok><mylist type="list"><item type="str">foo</item><item type="str">bar</item><item type="str">baz</item></mylist><mydict type="dict"><foo type="str">bar</foo><baz type="int">1</baz></mydict></root>

As you can see, the XML declaration has been omitted.

By default, dicttoxml includes a type attribute for each element. Starting in version 1.4, you can turn this off by passing an optional attr_type=False argument to the dicttoxml method.

Using our example:

>>> xml = dicttoxml.dicttoxml(obj, attr_type=False)
>>> print(xml)
<?xml version="1.0" encoding="UTF-8" ?><root><mydict><foo>bar</foo><baz>1</baz></mydict><mylist><item>foo</item><item>bar</item><item>baz</item></mylist><ok>true</ok></root>

As you can see, the only difference is that the type attributes are now absent.

By default, dicttoxml renders the XML element with an encoding="UTF-8" attribute. Starting in version 1.7.6, you can change the encoding by using the optional encoding argument to the dicttoxml method. For example, to render an XML file with encoding "ISO-8859-1", you would call:

>>> xml = dicttoxml.dicttoxml(obj, encoding="ISO-8859-1")

Or if you prefer, you can suppress the encoding attribute altogether by setting the optional include_encoding argument to False:

>>> xml = dicttoxml.dicttoxml(obj, include_encoding=False)

Again, by default, the include_encoding argument is set to True and the encoding argument is set to UTF-8.

Starting in version 1.1, you can set an optional ids parameter so that dicttoxml gives each element a unique id attribute.

With the ids flag on, the function generates a unique randomly-generated ID for each element based on the parent element in the form parent_unique. For list items, the id is in the form parent_unique_index.

Continuing with our example:

>>> xml_with_ids = dicttoxml.dicttoxml(obj, ids=True)
>>> print(parseString(xml_with_ids).toprettyxml())
<?xml version="1.0" ?>
<root>
        <mylist id="root_160980" type="list">
                <item id="mylist_609405_1" type="str">foo</item>
                <item id="mylist_609405_2" type="str">bar</item>
                <item id="mylist_609405_3" type="str">baz</item>
        </mylist>
        <mydict id="root_140407" type="dict">
                <foo id="mydict_260437" type="str">bar</foo>
                <baz id="mydict_111194" type="int">1</baz>
        </mydict>
        <ok id="root_612831" type="bool">true</ok>
</root>

Note that the default XML output remains the same as previous, so as not to break compatibility for existing uses.

Starting in version 1.3, dicttoxml accepts dict-like objects that are derived from the dict base class and treats them like dicts. For example:

>>> import collections
>>> dictlike = collections.OrderedDict({'foo': 1, 'bar': 2, 'baz': 3})
>>> xml = dicttoxml.dicttoxml(dictlike)
>>> print(xml)
<?xml version="1.0" encoding="UTF-8" ?><root><baz type="int">3</baz><foo type="int">1</foo><bar type="int">2</bar></root>

Also starting in version 1.3, dicttoxml accepts iterable objects and treats them like lists. For example:

>>> myiter = range(1,11)
>>> xml = dicttoxml.dicttoxml(myiter)
>>> print(xml)
<?xml version="1.0" encoding="UTF-8" ?><root><item type="int">1</item><item type="int">2</item><item type="int">3</item><item type="int">4</item><item type="int">5</item><item type="int">6</item><item type="int">7</item><item type="int">8</item><item type="int">9</item><item type="int">10</item></root>

As always, this remains compatible with arbitrary nesting of objects and types.

Starting in version 1.7, if you don't want item elements in a list to be called 'item', you can specify the element name using a function that takes the parent element name (i.e. the list name) as an argument.

>>> import dicttoxml
>>> obj = {u'mylist': [u'foo', u'bar', u'baz'], u'mydict': {u'foo': u'bar', u'baz': 1}, u'ok': True}
>>> my_item_func = lambda x: 'list_item'
>>> xml = dicttoxml.dicttoxml(obj, item_func=my_item_func)
>>> print(xml)
<?xml version="1.0" encoding="UTF-8" ?><root><mydict type="dict"><foo type="str">bar</foo><baz type="int">1</baz></mydict><mylist type="list"><list_item type="str">foo</list_item><list_item type="str">bar</list_item><list_item type="str">baz</list_item></mylist><ok type="bool">True</ok></root>

The benefit of taking the parent element name as an argument is that you can write the function to do something with it. Let's say you have an object with some lists of specific items:

>>> obj = {'shrubs': ['abelia', 'aralia', 'aucuba', 'azalea', 'bamboo', 'barberry', 'bluebeard', 'boxwood', 'camellia', 'dogwood', 'elderberry', 'enkianthus', 'firethorn', 'fuchsia', 'hazel', 'heath', 'heather', 'holly', 'honeysuckle', 'hydrangea', 'laurel', 'lilac', 'mock orange', 'rhododendron', 'rose', 'rose of sharon', 'rosemary', 'smokebush', 'spirea', 'sweetbox', 'viburnum', 'weigela', 'yucca'], 'trees': ['ash', 'aspen', 'birch', 'butternut', 'cedar', 'cottonwood', 'elm', 'fir', 'hawthorn', 'larch', 'locust', 'maple', 'oak', 'pine', 'spruce', 'sycamore', 'willow']}

You can define each item name to be the singular of its parent name by returning all but the last character.

>>> my_item_func = lambda x: x[:-1]
>>> xml = dicttoxml.dicttoxml(obj, item_func=my_item_func)
>>> print(xml)
<?xml version="1.0" encoding="UTF-8" ?><root><shrubs type="list"><shrub type="str">abelia</shrub><shrub type="str">aralia</shrub><shrub type="str">aucuba</shrub><shrub type="str">azalea</shrub><shrub type="str">bamboo</shrub><shrub type="str">barberry</shrub><shrub type="str">bluebeard</shrub><shrub type="str">boxwood</shrub><shrub type="str">camellia</shrub><shrub type="str">dogwood</shrub><shrub type="str">elderberry</shrub><shrub type="str">enkianthus</shrub><shrub type="str">firethorn</shrub><shrub type="str">fuchsia</shrub><shrub type="str">hazel</shrub><shrub type="str">heath</shrub><shrub type="str">heather</shrub><shrub type="str">holly</shrub><shrub type="str">honeysuckle</shrub><shrub type="str">hydrangea</shrub><shrub type="str">laurel</shrub><shrub type="str">lilac</shrub><shrub type="str">mock orange</shrub><shrub type="str">rhododendron</shrub><shrub type="str">rose</shrub><shrub type="str">rose of sharon</shrub><shrub type="str">rosemary</shrub><shrub type="str">smokebush</shrub><shrub type="str">spirea</shrub><shrub type="str">sweetbox</shrub><shrub type="str">viburnum</shrub><shrub type="str">weigela</shrub><shrub type="str">yucca</shrub></shrubs><trees type="list"><tree type="str">ash</tree><tree type="str">aspen</tree><tree type="str">birch</tree><tree type="str">butternut</tree><tree type="str">cedar</tree><tree type="str">cottonwood</tree><tree type="str">elm</tree><tree type="str">fir</tree><tree type="str">hawthorn</tree><tree type="str">larch</tree><tree type="str">locust</tree><tree type="str">maple</tree><tree type="str">oak</tree><tree type="str">pine</tree><tree type="str">spruce</tree><tree type="str">sycamore</tree><tree type="str">willow</tree></trees></root>

Of course, this can be combined with other optional arguments, like disabling type attributes or custom root element names.

Starting in version 1.7.1, you can wrap values in CDATA by setting the optional cdata argument to True.

>>> import dicttoxml
>>> obj = {u'mylist': [u'foo', u'bar', u'baz'], u'mydict': {u'foo': u'bar', u'baz': 1}, u'ok': True}
>>> xml = dicttoxml.dicttoxml(obj, cdata=True)
>>> print(xml)
<?xml version="1.0" encoding="UTF-8" ?><root><mydict type="dict"><foo type="str"><![CDATA[bar]]></foo><baz type="int"><![CDATA[1]]></baz></mydict><mylist type="list"><item type="str"><![CDATA[foo]]></item><item type="str"><![CDATA[bar]]></item><item type="str"><![CDATA[baz]]></item></mylist><ok type="bool"><![CDATA[True]]></ok></root>

If you do not set cdata to True, the default value is False and values are not wrapped.

By default, dicttoxml outputs the generated XML as a bytes object:

>>> xml = dicttoxml.dicttoxml(obj)
>>> type(xml).__name__
'bytes'

Starting in version 1.7.14, when you call the dicttoxml function, you can set an optional return_bytes argument to False to return a str object instead:

>>> xml = dicttoxml.dicttoxml(obj, return_bytes=False)
>>> type(xml).__name__
'str'

But by default, return_bytes is set to True.

As they say, Python comes with batteries included. You can easily syntax-check and pretty-print your XML using Python's xml.dom.minidom module.

Again, continuing with our example:

>>> from xml.dom.minidom import parseString
>>> dom = parseString(xml)
>>> print(dom.toprettyxml())
<?xml version="1.0" ?>
<root>
    <mylist type="list">
        <item type="str">foo</item>
        <item type="str">bar</item>
        <item type="str">baz</item>
    </mylist>
    <mydict type="dict">
        <foo type="str">bar</foo>
        <baz type="int">1</baz>
    </mydict>
    <ok type="bool">true</ok>
</root>

This makes the XML easier to read. If it is not well-formed, the xml parser will raise an exception.

You can enable debugging informationby calling the set_debug() method:

>>> import dicttoxml
>>> dicttoxml.set_debug(debug=True)
Debug mode is on. Events are logged at: dicttoxml.log
>>> xml = dicttoxml.dicttoxml(some_dict)

By default, debugging information is logged to dicttoxml.log, but you can change this:

>>> dicttoxml.set_debug(debug=True, filename='/path/to/some_other_filename.log')
Debug mode is on. Events are logged at: some_other_filename.log

To turn debug mode off, just call set_debug with an argument of False:

>>> dicttoxml.set_debug(debug=False)

If you encounter any errors in the code, please file an issue on github: https://github.com/quandyfactory/dicttoxml/issues.

• Author: Ryan McGreal
• Email: ryan@quandyfactory.com
• Repository: http://github.com/quandyfactory/dicttoxml

• Version: 1.7.16
• Release Date: 2022-12-23

• Release Date: 2022-12-23
• Changes:
  • Resolved issue #98 and issue #99 (big thanks to adrien-berchet for identifying the issue and proposing the solution)
  • Added python_requires=">=3.6" to setup.py

• Release Date: 2022-12-02
• Changes:
  • Fixed issue #43
  • Implemented issue #82
  • Small fixes to readme.md

• Release Date: 2022-12-02
• Changes:
  • Handle floating point keys as per issue #61
  • Option to return string instead of bytes as per issue #55
  • Reorganized readme.md to have a better flow.

• Release Date: 2022-11-29
• Changes:
  • Fixed issue #53, dicttoxml(None) and dicttoxml("name") break in 1.7
  • Fixed issue #96, update readme section on debugging

• Release Date: 2022-11-29
• Changes:
  • Fixed issue #95: changed project.toml to support Python 3.6+ and updated readme documentation.

• Release Date: 2022-11-28
• Changes:
  • Simplified solution to issue 94

• Release Date: 2022-11-28
• Changes:
  • Fixed issue #94

• Release Date: 2022-11-28
• Changes:
  • Fixed issue #67 and issue #81: Boolean values now export into XML in lowercase (true, false) instead of capitalized (True, False).

• Release Date: 2022-11-27
• Changes:
  • Fixed issue #58 and issue #60 on github: debug is turned off by default, and no longer prints "Debug mode is off" in the console.

• Release Date: 2022-11-26
• Changes:
  • Fixed issue #88 on github.
  • Made XML encoding attribute optional and editable as per issue 83 on github.

• Release Date: 2022-06-06
• Changes:
  • Fixed isue #91 on github.

• Release Date: 2016-07-08
• Changes:
  • Fixed bug #46 on github. Thanks to robbincatz for identifying and reporting the issue.

• Release Date: 2016-07-07
• Changes:
  • Updated README.markdown

• Release Date: 2016-07-07
• Changes:
  • XML-encodes values to avoid XML injection. Big thanks to thomaskonrad on Github, via issue #41.

• Release Date: 2016-07-06
• Changes:
  • Added ability to wrap values with CDATA. Big thanks to LeviTaule on Github, via pull request #45.

• Release Date: 2016-06-13
• Changes:
  • First of all, sorry for such a log delay between releases! I have not been a responsible steward of this project and I aim to change that from now on. This is the first in a series of updates I will be pushing over the next couple of months to get caught up on the backlog of issues and pull requests.
  • Added ability to customize list and dict item names via a function argument passed into the dicttoxml() function. Customizeable item name function takes the item's parent element as an argument. Big thanks to viktor-zireael on Github, via pull request #40.
  • Updated code style to more closely follow PEP8.

• Release Date: 2015-04-09
• Changes:
  • PyPi does not want to upload version 1.6.5. It's returning an Upload failed (500): Internal Server Error message when I try to upload the code. I'm incrementing the version by one and reinstalling it to see if that fixes the issue.

• Release Date: 2015-04-09
• Changes:
  • Fixed issue #37, elements with boolean values were getting a "number" type attribute. The issue was that isinstance(True, numbers.Number) returns True. I modified the get_xml_type() function to test for boolean before testing for numbers.Number. Thanks to badsequel for identifying and reporting the issue.

• Release Date: 2015-03-11
• Changes:
  • Fixed issue #36, logging was throwing an UnicodeDecodeError on non-ASCII characters in dictionary values. Thanks to beef9999 for identifying and reporting the issue.

• Release Date: 2015-03-05
• Changes:
  • Updated README.markdown to reflect changes made in v. 1.6.2.

• Release Date: 2015-03-05
• Changes:
  • Fixed issue #35, dicttoxml fails to identify a decimal.Decimal as a number. This is done by replacing type(val).__name__ in ('int', 'long') with the more generic isinstance(val, number.Number). Thanks to jmagnusson for finding and fixing the error.

• Release Date: 2015-03-05
• Changes:
  • Merged pull request #34, fix misleading TypeError in convert_dict(). Thanks to jmagnusson for finding and fixing the error.

• Release Date: 2015-02-23
• Changes:
  • Fixed issue #32, duplication in test for list-like data types.

• Release Date: 2015-02-23
• Changes:
  • Merged pull request #33 to replace invocations of logging with LOG. Thanks to mfriedenhagen for identifying the issue with the logger, and to seyhuns for supplying a pull request that could be merged automatically.

• Release Date: 2015-01-06
• Changes:
  • Fixed issue #30 via pull request #31. Thanks to isaac-councill for identifying the issue and providing a fix.

• Release Date: 2014-12-09
• Changes:
  • Fixed issue #29. Thanks to birdsarah for identifying this performance issue and providing a fix.

• Release Date: 2014-08-18
• Changes:
  • Fixed issue #24. Thanks to gdude2002 for identifying the issue.
  • Abstracted all XML validity tests to a single function make_valid_xml_name(key, attr)

• Release Date: 2014-06-16
• Changes:
  • Fixed issue #21. Thanks to lichenbo for identifying the issue and providing a fix.
  • Abstracted setting XML type attribute into a function, get_xml_type().
  • Standardized variable names inside functions (e.g. k -> key, v -> val).
  • Cleaned up README so it works as both Markdown (for github) and ReStructuredText (for PyPi)

• Release Date: 2014-06-03
• Changes:
  • Fixed issue #20. Thanks to lichenbo for identifying the issue and providing a fix.

• Release Date: 2014-06-08
• Changes:
  • Minor updates to README.markdown

• Release Date: 2014-06-03
• Changes:
  • Minor updates to README.markdown

• Release Date: 2014-06-03
• Changes:
  • Minor updates to README.markdown

• Release Date: 2014-06-03
• Changes:
  • Added ability to set a custom root element, as per issue #18 by murielsilveira.

• Release Date: 2014-06-03
• Changes:
  • Element type attribute made optional via pull request from [gauravub](https://github.com/gauravub] to resolve issue #17.

• Release Date: 2014-04-21
• Changes:
  • Updated MANIFEST.in and setup.py so the licence and readme are properly included in the distribution.

• Release Date: 2014-04-21
• Changes:
  • Added MANIFEST.in to include the LICENCE.txt and README.markdown files in the distribution, as per issue #15.

• Release Date: 2014-04-14
• Changes:
  • dicttoxml() accepts [None] as a parameter and returns a valid XML object, as per issue #13.

• Release Date: 2014-04-14
• Changes:
  • dicttoxml() now accepts None as a parameter and returns a valid XML object, as per issue #13.

• Release Date: 2014-04-14
• Changes:
  • Automatically converts spaces in key names to underscores, as per issue #12.

• Release Date: 2014-04-14
• Changes:
  • Added convert_none() function to convert a null value into XML
  • Added key_is_valid_xml() function to test if a key is valid XML
  • Updated convert_kv(), convert_bool() and convert_none() functions to test whether the key is a valid XML name and, if it is not, to render it as <key name="{invalidname}">value</key>. This addresses issue 10.

• Release Date: 2013-07-12
• Changes:
  • Updated README to note support for dict-like and iterable objects.

• Release Date: 2013-07-12
• Changes:
  • changed test for dict type from type(x)=dict to isinstance(x,dict) to include dict-like subclases derived from dict, as per issue 9.
  • Added test for isinstance(x,collections.Iterable) to test for list, set, tuple to accommodate iterable objects, as per issue 9.

• Release Date: 2013-07-11
• Changes:
  • Fixed typo in convert_list() exception raise as per issue 8.

• Release Date: 2013-05-06
• Changes:
  • Renamed github repo from dict2xml to dicttoxml to match PyPI name.

• Release Date: 2013-05-06
• Changes:
  • Fixed README.markdown

• Release Date: 2013-05-06
• Changes:
  • Added an optional ids argument to give each element a unique, randomly generated id attribute.
  • All elements now inlcude a type attribute.
  • Updated readme with more examples and Python 3 compatible syntax.
  • Thanks to cpetz for suggesting this feature.

• Release Date: 2013-03-04
• Changes:
  • Replaced debug function with logging module.
  • Converted code to work in Python 2.6+ and Python 3.
  • Fixed unresolved isoformat reference in convert_list.
  • Bug thanks to regisd for forking code and making several important fixes!

• Release Date: 2013-03-03
• Changes:
  • Merged pull request from regisd to fix issue #5, in which special XML characters were not being escaped properly.

• Release Date: 2013-02-27
• Changes:
  • Added support for tuples.

• Release Date: 2013-02-23
• Changes:
  • Changed name to dicttoxml and published to the Python Package Index (PyPI).

• Release Date: 2012-09-12
• Changes:
  • Fixed issue #4 - thanks to PaulMdx for finding it and suggesting a fix.

• Release Date: 2012-07-13
• Changes:
  • Merged pull request from 0902horn on github to escape special XML characters.

• Release Date: 2012-02-28
• Changes:
  • Added support for datetime objects (converts them into ISO format strings) and sets (converts them into lists).
  • Fixed bug 2 by raising an exception on unsupported data types.

• Release Date: 2012-01-26
• Changes:
  • Added optional root argument (default True) on whether to wrap the generated XML in an XML declaration and a root element.
  • Added ability to convert a root object of other data types - int, float, str, unicode, list - as well as dict.
  • Corrected license attribute in setup.py.
  • Renamed notify() function to debug_notify() and made it more comprehensive.

• Release Date: 2012-01-24
• Changes:
  • Fixed inconsistent str/string attributes.

• Release Date: 2012-01-24
• Changes:
  • Fixed bug in list items.
  • Added element attribute with data type.

• Release Date: 2012-01-24
• Changes:
  • First commit.

Copyright 2012 by Ryan McGreal.

Released under the GNU General Public Licence, Version 2:
http://www.gnu.org/licenses/old-licenses/gpl-2.0.html