open-contracting / extension_registry.py Goto Github PK

Related open-contracting-archive/extensions-data-collector#4 (comment)

Once this is available, an argument can be provided to ExtensionRegistry to read all data locally.

Right now, it downloads the ZIP file from GitHub, which will be the English schema.

For example, in open-contracting-extensions/public-private-partnerships#243, initiationType.csv was removed from ocds_ppp_extension, but the file wasn't removed by build-profile.py in public-private-partnerships.

e.g. if we want to generate an ad-hoc 'profile' with which to validate data (in CoVE) or with which to determine which quality checks to run (in Data Quality Tool, e.g. finding all Period objects in the schema and then running period-specific checks).

cc @kindly as FYI

CoVE might do some things that ProfileBuilder doesn't (yet).

Ideally, all projects would use the same library.

I think the two changes needed are to:

• Lazy-load the registry
• Ensure that local file paths can be provided as a list to extension_versions parameter

If this is done, then, e.g. Kingfisher can have a post-scrape step to download all unique extensions referenced by packages. The ProfileBuilder can then use these to generate an ad-hoc 'profile' (#27), which can be made available to other steps (e.g. the check step).

See open-contracting/spoonbill#185 (comment)

https://ocdsextensionregistry.readthedocs.io/en/latest/translation.html

The ocds-extensions-translations repo can still have po files for all versions, but we don't need to add old versions to Transifex, since almost no users (per Google Analytics) access them.