Warning

This extension is now deprecated.

Due to the API changes Twitter/X introduced in 2023, this extension no longer works and will not be fixed. There will not be any further releases, new features, or bugfixes.

A CKAN extension that enables users to post a tweet every time a dataset is created or updated.

This extension connects a CKAN instance to a Twitter account so that when a dataset is updated or created (i.e. the after_update hook is called), the user has the option to post a tweet about the activity.

Path variables used below:

• $INSTALL_FOLDER (i.e. where CKAN is installed), e.g. /usr/lib/ckan/default
• $CONFIG_FILE, e.g. /etc/ckan/default/development.ini

pip install ckanext-twitter

1. Clone the repository into the src folder:

   cd $INSTALL_FOLDER/src
   git clone https://github.com/NaturalHistoryMuseum/ckanext-twitter.git

2. Activate the virtual env:

   . $INSTALL_FOLDER/bin/activate

3. Install via pip:

   pip install $INSTALL_FOLDER/src/ckanext-twitter

Installing from a pyproject.toml in editable mode (i.e. pip install -e) requires setuptools>=64; however, CKAN 2.9 requires setuptools==44.1.0. See our CKAN fork for a version of v2.9 that uses an updated setuptools if this functionality is something you need.

1. Add 'twitter' to the list of plugins in your $CONFIG_FILE:

   ckan.plugins = ... twitter

2. Add a 'tweet' block to read_base.html:

   {% block tweet %}
   {{ super() }}
   {% endblock %}

3. Optionally, override the styling of the block by creating an ajax_snippets/edit_tweet.html file.

These are the options that can be specified in your .ini config file. The only required options are the twitter credentials. Everything else has a sensible default set.

All of these can be obtained by creating a single-user app at apps.twitter.com. They can be found on the "keys and access tokens" tab when viewing your app.

Tweets are generated using Jinja2 and use tokens derived from the package dictionary. See Usage for more detail.

If your config is created dynamically using Jinja2, you will have to wrap any custom template in {% raw %}{% endraw %} tags and add a newline after it, e.g.:

ckanext.twitter.new = {% raw %}{{ title }} by {{ author }} ({{ records }} records) has just been published!{% endraw %}

ckanext.twitter.consumer_key = {{ twitter.consumer_key }}
ckanext.twitter...

Token values for the tweet templates will come from a simplified package dictionary. In these, any collection values (i.e. lists and dictionaries) have been replaced with the number of items, the author list has been significantly shortened, and any long strings will be shortened to fit into the tweet character limit (currently set at 140).

For example, if the package dictionary is:

{
  'author': 'Dippy Diplodocus, Sophie Stegosaurus',
  'author_email': None,
  'dataset_category': ['Citizen Science'],
  'doi': 'DOI_VALUE',
  'license_title': 'Creative Commons Attribution',
  'organization': {'description': '', 'name': 'nhm', 'is_organization': True, 'state': 'active'},
  'resources': [
                  {'mimetype': 'image/jpeg', 'name': 'resource_1.jpg', 'format': 'JPEG'},
                  {'mimetype': 'image/jpeg', 'name': 'resource_2.jpg', 'format': 'JPEG'}
               ],
  'title': 'Dataset Name'
}

Then the tokenised dictionary would be:

{
  'author': 'Diplodocus et al.',  # just the surname of the first author
  # author_email was None so it's excluded
  'dataset_category': 1,  # lists are counted
  'doi': 'DOI_VALUE',  # simple string values stay the same
  'license_title': 'Creative Commons Attribution',
  'organization': 4,  # dicts are also counted
  'resources': 2,
  'title': 'Dataset Name'
}

And if you had defined the tweet template as:

New dataset: "{{ title }}" by {{ author }} ({{ resources }} resources).

Your tweet would then read:

New dataset: "Dataset Name" by Diplodocus et al. (2 resources)

There is a Docker compose configuration available in this repository to make it easier to run tests. The ckan image uses the Dockerfile in the docker/ folder.

To run the tests against ckan 2.9.x on Python3:

1. Build the required images:

   docker-compose build

2. Then run the tests. The root of the repository is mounted into the ckan container as a volume by the Docker compose configuration, so you should only need to rebuild the ckan image if you change the extension's dependencies.

   docker-compose run ckan

Note that the tests shouldn't make any calls to Twitter's API and won't post any tweets.