Contributing

Documentation

Superdesk documentation is written using rst format and generated via Sphinx. It’s organized by topic, using autodoc as much as possible to include docstrings from python code.

When working on docs, you can use live preview. In docs folder with virtualenv enabled run:

$ make livehtml

This will build docs and run a server on localhost:8000. It will refresh as you modify documentation, but not when you modify python docstrings, in order to see some changes done there you still have to make some changes in rst files.

Once you make a PR and it gets merged, you will see updated docs on superdesk.readthedocs.io.

Updating docs

Documentation should be added/updated together with code changes in a single PR to superdesk-core repository. There can be also PRs with only documentation.

New topic/module

To add a new topic or module docs, you create a file eg. foo.rst in docs folder and then you have to add it to index.rst toctree. This will make it appear in table of contents in both sidebar and on homepage.

Docs conventions

Again - we should use autodocs as much as possible so that documentation is close to code and thus should get updated with it. Thus to document a class or function, use autoclass and autofunction:

.. autoclass:: apps.publish.content.common.BasePublishService
    :members:

.. autofunction:: superdesk.publish.transmit

If you want to document multiple classes/functions from same module, you should use automodule or module first:

.. module:: superdesk.io.feed_parsers

.. autoclass:: ANPAFeedParser
.. autoclass:: IPTC7901FeedParser

If you want to document whole module with all its members, you can just use automodule:

.. automodule:: superdesk.io.feed_parsers
    :members:

This will document all public members from the module which have a docstring.

You can integrate UML diagrams using PlantUML syntax, with:

.. uml::

    [plantuml diagram]

Formatter

This is a short how-to to create a formatter.

Generic formatter are put in superdesk-core repository, you’ll find them in superdesk/publish/formatters directory. Sometimes, we need to do custom version of formatter, in this case they are put in a fork of superdesk repository, in server/[fork_dir]/formatters.

To create a new formatter, you can either subclass an existing one, or start fresh from superdesk.publish.formatters.Formatter

class superdesk.publish.formatters.Formatter

Base Formatter class for all types of Formatters like News ML 1.2, News ML G2, NITF, etc.

Checks if the article has any Public Service Announcements and if available appends each of them to the body.

Returns:body with public service announcements.

Checks if the article has the legal flag on and adds ‘Legal:’ to the slugline

Parameters:
  • article – article having the slugline
  • truncate – truncates the slugline to 24 characters
Returns:

updated slugline

can_format(format_type, article)

Test if formatter can format for given article.

export(article, subscriber, codes=None)

Formats the article and returns the output string for export

format(article, subscriber, codes=None)

Formats the article and returns the transformed string

map_html_to_xml(element, html)

Map the html text tags to xml

Parameters:
  • element (etree.Element) – The xml element to populate
  • html (str) – the html to parse the text from
Returns:

The most important method here is format which is the one called with the article to format. This method must return a list of 2 elements tuples with a publish sequence number and the formatted output. You can get publish sequence number using subscribers service:

pub_seq_num = superdesk.get_resource_service('subscribers').generate_sequence_number(subscriber)

You should wrap your format method in a try..except, and raise a FormatterError if any problem arise.

class superdesk.errors.FormatterError(code, exception, destination=None)

Register formatter

To use your formatter, you need to register it. This is done simply by importing your new module in superdesk.publish.formatters.__init__. Here we do it for NITF:

from .nitf_formatter import NITFFormatter  # NOQA

Note the # NOQA which will avoid troubles with flake8 (the module is imported but not used immediately). The registration is done automatically thanks to the Formatter class.