NewsML-G2 - Python implementation of the NewsML-G2 standard

NewsML-G2 is an open standard created by the International Press Telecommunications Council to share news content. See http://www.newsml-g2.org/

This module is a part-implementation of the standard in Python. Currently it reads itemMeta and contentMeta blocks, catalogs and metadata objects from NewsML-G2 XML files and outputs Python objects.

Currently built for Python 3 only - please let us know if you require Python 2 support.

Installation

Installing from PyPI:

pip install newsmlg2

Reading NewsML-G2 files

Example:

import NewsMLG2

# load NewsML-G2 from a file and print the parsed version
g2doc = NewsMLG2.NewsMLG2Document("test-newsmlg2-file.xml")
print(g2doc.get_item())

# load NewsML-G2 from a string
g2doc = NewsMLG2.NewsMLG2Document(
b"""<?xml version="1.0" encoding="UTF-8"?>
<newsItem
    xmlns="http://iptc.org/std/nar/2006-10-01/"
    guid="simplest-test"
    standard="NewsML-G2"
    standardversion="2.36"
    conformance="power"
    version="1"
    xml:lang="en-GB">
    <catalogRef href="http://www.iptc.org/std/catalog/catalog.IPTC-G2-Standards_41.xml" />
    <itemMeta>
        <itemClass qcode="ninat:text" />
        <provider qcode="nprov:IPTC" />
        <versionCreated>2025-09-29T12:00:00+03:00</versionCreated>
    </itemMeta>
    <contentSet>
        <inlineXML contenttype="application/nitf+xml">
        </inlineXML>
    </contentSet>
</newsItem>
""")

# get the newsItem from the parsed object
newsitem = g2doc.get_item()
# test various elements and attributes using our shortcut dot syntax
assert newsitem.guid == 'simplest-test'
assert newsitem.standard == 'NewsML-G2'
assert newsitem.standardversion == '2.36'
assert newsitem.conformance == 'power'

itemmeta = newsitem.itemmeta
# you can choose whether to use qcodes or URIs, we do the conversion for you
# using the catalog declared in the NewsML-G2 file
assert itemmeta.itemclass.qcode == 'ninat:text'
assert NewsMLG2.qcode_to_uri(itemmeta.itemclass.qcode) == 'http://cv.iptc.org/newscodes/ninature/text'
assert itemmeta.provider.qcode == 'nprov:IPTC'
assert NewsMLG2.qcode_to_uri(itemmeta.provider.qcode) == 'http://cv.iptc.org/newscodes/newsprovider/IPTC'
# Elements that contain a simple text string can be read with str(class)
assert str(itemmeta.versioncreated) == '2025-09-29T12:00:00+03:00'

etc...

Creating NewsML-G2 files using Python code

There are a few points to note when creating NewsML-G2 directly in Python code (as opposed to parsing a string containing XML).

• Elements with multiple values (such as multiple elements) must be created individually and then added to their parent through array assignment. So you should create child elements and then add them as an array of values.

See the below example which adds multiple <broader> elements inside the located element:

g2doc = NewsMLG2.NewsMLG2Document()
newsitem = NewsMLG2.NewsItem()
newsitem.guid = 'test-guid'
newsitem.xml_lang = 'en-GB'
itemmeta = NewsMLG2.ItemMeta()
itemmeta.itemclass.qcode = "ninat:text"
itemmeta.provider.qcode = "nprov:IPTC"
itemmeta.versioncreated = "2025-09-29T12:00:00+03:00"
newsitem.itemmeta = itemmeta
contentmeta = NewsMLG2.NewsItemContentMeta()
contentmeta.contentcreated = '2008-11-05T19:04:00-08:00'
located = NewsMLG2.Located()
located.type = 'cptype:city'
located.qcode = 'city:345678'
located.name = 'Berlin'
contentmeta.located = located
located = NewsMLG2.Located()
located.type = 'cptype:city'
located.qcode = 'city:345678'
located.name = 'Berlin'
contentmeta.located = located
digsrctype = NewsMLG2.DigitalSourceType()
digsrctype.uri = 'http://cv.iptc.org/newscodes/digitalsourcetype/trainedAlgorithmicMedia'
contentmeta.digitalsourcetype = digsrctype
broader1 = NewsMLG2.Broader()
broader1.type = 'cptype:statprov'
broader1.qcode = 'state:2365'
broader1.name = 'Berlin'
broader2 = NewsMLG2.Broader()
broader2.type = 'cptype:country'
broader2.qcode = 'iso3166-1a2:DE'
broader2.name = 'Germany'
contentmeta.located.broader = [broader1, broader2]
creator = NewsMLG2.Creator()
creator.qcode = 'codesource:DEZDF'
creator.name = 'Zweites Deutsches Fernsehen'
# This implements
# contentmeta.creator.organisationdetails.location.name = 'MAINZ'
# we have to make each item separately.
orgdetails = NewsMLG2.OrganisationDetails()
orglocation = NewsMLG2.OrganisationLocation()
orglocation.name = 'MAINZ'
orgdetails.location = orglocation
creator.organisationdetails = orgdetails
contentmeta.creator = creator
newsitem.contentmeta = contentmeta
g2doc.set_item(newsitem)

output_newsitem = g2doc.get_item()
assert newsitem.guid == 'test-guid'
assert newsitem.standard == 'NewsML-G2'
assert newsitem.standardversion == '2.36'
assert newsitem.conformance == 'power'
assert newsitem.version == '1'
assert newsitem.xml_lang == 'en-GB'

output_xml = g2doc.to_xml_string()

assert output_xml == (
    "<?xml version='1.0' encoding='utf-8'?>\n"
    '<newsItem xmlns="http://iptc.org/std/nar/2006-10-01/" xmlns:nitf="http://iptc.org/std/NITF/2006-10-18/" xml:lang="en-GB" standard="NewsML-G2" standardversion="2.36" conformance="power" guid="test-guid" version="1">\n'
    '  <itemMeta>\n'
    '    <itemClass qcode="ninat:text"/>\n'
    '    <provider qcode="nprov:IPTC"/>\n'
    '    <versionCreated>2025-09-29T12:00:00+03:00</versionCreated>\n'
    '  </itemMeta>\n'
    '  <contentMeta>\n'
    '    <contentCreated>2008-11-05T19:04:00-08:00</contentCreated>\n'
    '    <located qcode="city:345678" type="cptype:city">\n'
    '      <name>Berlin</name>\n'
    '      <broader qcode="state:2365" type="cptype:statprov">\n'
    '        <name>Berlin</name>\n'
    '      </broader>\n'
    '      <broader qcode="iso3166-1a2:DE" type="cptype:country">\n'
    '        <name>Germany</name>\n'
    '      </broader>\n'
    '    </located>\n'
    '    <digitalSourceType uri="http://cv.iptc.org/newscodes/digitalsourcetype/trainedAlgorithmicMedia"/>\n'
    '    <creator qcode="codesource:DEZDF">\n'
    '      <name>Zweites Deutsches Fernsehen</name>\n'
    '      <organisationDetails>\n'
    '        <location>\n'
    '          <name>MAINZ</name>\n'
    '        </location>\n'
    '      </organisationDetails>\n'
    '    </creator>\n'
    '  </contentMeta>\n'
    '</newsItem>\n')

Testing

A unit test library is included.

Run it with:

pytest

Test coverage can be measured with the coverage.py package:

pip install coverage
coverage run --source NewsMLG2 -m pytest 
coverage report

Release notes

• 0.1 - First release, pinned to Python 3 only (use pip >9.0 to ensure pip's Python version requirement works properly)
• 0.2 - Can now read and write NewsML-G2 from code - NewsMessage and PlanningItem not yet implemented. Probably quite a few bugs.
• 0.3 - Changed from automatically converting between URIs and QCodes to providing helper functions uri_to_qcode() and qcode_to_uri()
• 0.4 - Added catalog v37 and v38. Added PlanningItem support. Fixed bugs. Improved magic function support to help hasattr() and more on NewsML-G2 objects.
• 0.5 - Now has 100% unit test coverage. Fixed more bugs. Implemented changes up to NewsML-G2 schema version v2.32.
• 0.6 - Implemented NewsMessage and Events (EventsML-G2). Adding arrays using code (as opposed to parsing an XML string/file) now works. Almost ready to go to 1.0.
• 0.7 - Fixed a bug whereby IPTC catalog files were not included in the distribution package.
• 0.8 - Switched to latest setuptools packaging configuration.
• 0.9 - Fixed a bug with packaging in the previous version. Added support for XML enumerations. Updated catalog cache to include latest versions.
• 0.10 - Fixed more packaging bugs.
• 1.0 - First stable release. Added xs:any support and roundtrip tests.
• 1.1 - Change NewsMLG2Document() init so it can accept either filename or string, as per the above examples. Changed to_xml() to output etree XML elements and added to_xml_string() to output XML as a string. Added tests to ensure that the README examples work.
• 1.2 - Updated to NewsML-G2 2.36, adding aiInfo and its children. Fixed some small bugs. Bundled all IPTC G2 Standards catalogs from v1 to v42 plus the IPTC Sports catalogs v1 and v2.