Support ZIMing local Kolibri data created by a dry-run of ricecooker

[benoit74]

Many (most ?) Kolibri channels are populated by ricecooker which creates everything needed locally and then uploads this to a Kolibri Studio Channel

The scraper then retrieves this Studio Channel and creates a ZIM.

While convenient when the Studio Channel is well maintained, it proves to be a burden when we struggle to get Studio Channel updated (like we face with openzim/zim-requests#729).

We might want to consider to:

• run ricecooker in "dry-run" mode, where files are created locally but not uploaded to a Studio Channel
• directly process these files with the scraper to create a ZIM

This change might be quite straightforward since what we download from the Studio Channel is maybe very close to what we would get locally from rice-cooker. And we would continue to benefit from the rest of the scraper enhancements (ZIM UI, ...).

[rgaudin]

I believe the collection is just a SQLite DB with files named based on IDs in the DB. Why not adding a script to either:

• build a collection similar to that of studio from that dry-run folder. Might already exists.
• serve (using python's bundled http server) that dry-run folder in a way that mimics the studio

This would eliminate the need for scraper change. Running it with just a env change for the studio URL

[benoit74]

Yup, to be investigated, good idea