Get the URL and the repository from browser-specs

[deniak]

That PR leverages browser-specs to get the URL and the repo for each spec.

If a spec is unknown, an error is returned and if the shortname used is an old one, a warning is issued. The idea is to deal with this manually.

@tidoust mentioned that the tests property in browser-specs should also give us the right WPT path but in some cases, e.g. speculation-rules, I'm not sure how they are dealing with the different features. In any case, it's good to have a fallback mechanism until WPT has the data we need.

Fix #33

[tidoust]

@tidoust mentioned that the tests property in browser-specs should also give us the right WPT path but in some cases, e.g. speculation-rules, I'm not sure how they are dealing with the different features.

In theory, there should be a META.yml file in appropriate subfolders. The tests property in browser-specs both has a testPaths key that lists WPT folders to include and an excludePaths key that lists subfolders within these folders that should not be taken into account.

The information in WPT does not seem up-to-date for the speculation-rules folder. At a minimum, a META.yml file is missing from speculation-rules/prefetch. There are likely other updates needed there in theory because the initial speculation-rules spec got partially merged into HTML.

[tidoust]

The information in WPT does not seem up-to-date for the speculation-rules folder. At a minimum, a META.yml file is missing from speculation-rules/prefetch. There are likely other updates needed there in theory because the initial speculation-rules spec got partially merged into HTML.

I created web-platform-tests/wpt#63327 to update the information in WPT.

[kfranqueiro]

Is there a reason to filter this way, vs. referencing resolvedSpecs?

[deniak]

resolvedSpecs is a subset of specs (with additional properties). So if for some reasons, a spec listed in specs.json is dropped out from browser-specs, it will no longer be in resolvedSpecs and hence, will not be processed.
Filtering specs will ensure we keep the existing data available until specs.json is cleaned up. index.js will report an error (l.130) in such case and a manual cleanup is expected.