ckanext-zenodo

Stores a dataset's large or non-tabular files on Zenodo instead of in CKAN's own storage, and publishes them there - with a DOI, in a Zenodo community - when the dataset is made public. The dataset stays a normal, searchable CKAN dataset: each Zenodo file appears in it as a resource linking to Zenodo, with a preview for images, audio, video and PDF. Built for the FAIR3R catalogue, whose FDF metadata it maps into the Zenodo record.

How it works

• Routing: a file goes to Zenodo when it's larger than ckan.max_resource_size, or in a format the data viewer can't read (not in ckanext.xloader.formats), or when the user switches "Zenodo" on. Everything else stays in CKAN.
• Upload: the file is streamed through the CKAN server into a draft Zenodo deposition kept for the dataset, by one shared Zenodo account. Nothing is public yet.
• Publish: when the dataset is active and public, a background job writes the metadata, publishes the deposition (Zenodo mints a DOI), files it into the community and creates the CKAN resources.
• Later changes: adding, replacing or deleting a file publishes a new version of the same record. Resources link to the record's concept DOI, which always resolves to the latest version.
• Dataset deleted: unpublished Zenodo files are deleted with it; published ones stay on Zenodo (the user is told so).

The full specification is in docs/: architecture, every workflow step by step, metadata mapping, security, configuration and operations, and the Zenodo API behaviour it relies on.

Requirements

• CKAN 2.12 with its Redis and background worker (ckan jobs worker) running - publishing and deletions run as background jobs.
• A Zenodo account and a personal access token (Zenodo -> Applications -> Personal access tokens) allowed to create and publish depositions (deposit:write, deposit:actions). Use sandbox.zenodo.org for testing: it's a separate site, with its own account, token and community.
• A Zenodo community, of which that account is owner, manager or curator (needed to accept each record into it).
• Recommended: ckanext-fair3r (the upload form that routes files here, and the FDF metadata) and ckanext-doi (link from the Zenodo record back to the dataset's DOI). Without ckanext-fair3r, nothing in CKAN's standard interface sends files here.

Installation

1. Install the package in CKAN's virtualenv:

   pip install ckanext-zenodo
   

2. Enable the plugin, with CKAN's view plugins used for the previews, in ckan.ini:

   ckan.plugins = ... zenodo image_view audio_view video_view pdf_view
   

3. Configure the Zenodo account (sandbox here; for zenodo.org set test_mode = false and ckanext.zenodo.access_token instead):

   ckanext.zenodo.test_mode = true
   ckanext.zenodo.sandbox_access_token = <token>
   ckanext.zenodo.community_identifier = fair3r
   

   Every other setting has a default - see docs/operations.md.

4. Create the table:

   ckan -c /path/to/ckan.ini db upgrade -p zenodo
   

5. Retry failed publishes every 30 minutes, as the CKAN user (cron):

   */30 * * * * /path/to/venv/bin/ckan -c /path/to/ckan.ini zenodo retry-pending
   

6. Let large uploads through: the reverse proxy's body size limit for /zenodo/upload/ must allow ckanext.zenodo.max_file_size_bytes (10 GB by default), and the server needs that much free disk on top of ckanext.zenodo.disk_safety_margin_bytes (5 GB) - each upload is written to a temporary file before being sent on.

7. Restart CKAN and its worker, then check the setup against Zenodo:

   ckan -c /path/to/ckan.ini zenodo check
   

   It verifies the token, its access to depositions, the community and the account's role in it, and exits 1 on the first problem.

Commands

Relationship to ckanext-fair3r

Not a declared dependency, but the reason this plugin exists:

• ckanext-fair3r's resource form decides, in the browser, which files go to Zenodo and sends them to this plugin's /zenodo/upload/<dataset> endpoint (with the "Zenodo" switch and the live capacity badge). It also provides the "Publish" button and "Download all", which uses Zenodo's own ZIP for Zenodo files.
• This plugin converts the dataset's FDF metadata with ckanext-fair3r's DataCite converter (optional import). Without it, a record is still published, with minimal metadata (title, tags, licence, submitter).

Tests

docker exec -u ckan -it ckan-app pytest --ckan-ini=/plugins/ckanext-zenodo/test.ini /plugins/ckanext-zenodo/ckanext/zenodo/tests

Always pass --ckan-ini: the test fixtures empty the database they run against, and refuse to run unless its name contains test. The test.ini at the repository root is for the DEV container; ckanext/zenodo/tests/test.ini, shipped in the package, is the one the integration server uses - keep their plugin lists in sync.

License

Released under the GNU Affero General Public License (AGPL) v3.0. See the file LICENSE for details.