Store license list in database

[davidread]

It would be useful to add licenses using an API call - this is not currently possible.

The use case I have is: data.gov.uk harvests lots of datasets from ISO19139 format which contains the license expressed in free text. It would be useful to add these as ckan licenses at the harvest. What we've ended up doing is storing the licence in an extra field, which is a bit non-standard and means we have to customize everything that touches licences (e.g. testing for openness), let alone end-users using this via the API, sometimes in comparison with other CKANs.

A second use case is that a sysadmin can change the list of licences by using the web interface, rather than the command-line.

Currently the list of licenses is initialized from either a json file or a basic list, stored in code. https://github.com/ckan/ckan/blob/master/ckan/model/license.py#L43-68

Details of proposal:

• 'license' table has columns: id, name, title, od_compliance, od_url (on opendefinition.org), home_url (optional, e.g. Creative Commons' URL). Drop info about family, maintainer, status, osi_compliance - CKAN doesn't need to know these and they can be got from opendefinition.org if necessary.
• When setting up a new CKAN you do "paster db init" and this would add just one licence - CC-BY-4.0 (to encourage this).
• Existing CKAN databases would run "paster db upgrade" and the migration would add to the database all licenses currently in use by existing datasets.
• Action APIs added to search, add & delete licences. This allows you to pipe in an existing JSON file or a licence from opendefinition.org.
• We compromise translated sites a bit - we'd no longer have the Chinese names for the basic licences - Creative Commons, ODL and ODbL - they'd have to add these using the API if they want them. However we want to discourage them from offering multiple licences (https://github.com/ckan/ckan/pull/2002) - one should be enough for many CKAN sites. (Translatability is why Toby changed our default list of licences from JSON to code: https://github.com/ckan/ckan/commit/3ee86db592d5717ac426b46b91f16c4bc3efc51a )

[iaroslav13]

I am working on this changes now.

Basically I’ve already implemented changes described in issue description. But there are some things I am not sure about:

• Columns of license table. I don’t agree with the proposed structure: id, name, title, od_compliance, od_url (on opendefinition.org), home_url (optional, e.g. Creative Commons' URL).

In my opinion we need to keep almost the same structure that was hardcoded licenses but to drop info about family, maintainer, status, osi_compliance as suggested

I am proposing the following structure that I already implemented: id text NOT NULL, title text NOT NULL, is_okd_compliant boolean, is_generic boolean, url text, home_url text, status license_status NOT NULL id should be unique 'license_status' type possible values are ‘active' and ‘deleted'

• I created views/forms that allows users to view the list of licenses, add new licenses and edit existing ones. But it is not clear where the link to License list should be placed. Currently I placed it on the Administration page (/ckan-admin) as a tab. I don’t think this is a good place for the link and I’d like to find a better place for it.

In my opinion changes described in the description are really important and therefore I’d like to get responses to my questions from dev team. Also if it is possible I’d like to add these changes to upcoming 2.4 release.

You can find my changes here: https://github.com/iaroslav13/ckan/commit/a06742fbf5ae39496283a9feece847fb3dfe39e8

[wardi]

When specifying licenses in a json file we can add arbitrary extra fields the the licenses object. We use this to store translations of some of the core fields.

I ask that we add a place for extras in the database and have them populate the license returned similar to how it's done in resources (all extras at the same level as the core fields)

[iaroslav13]

I made changes so that custom fields can be added to license records via API. See my pull request: https://github.com/ckan/ckan/pull/2452