Change "next_run" to "best_run" and refactor how next_start_date is calculated

[mbertrand]

What are the relevant tickets?

• Backend part of https://github.com/mitodl/hq/issues/8818

Description (What does it do?)

Replaces LearningResource.next_run with LearningResource.best_run and changes the criteria for selecting that run. This way, an enrollable run that is currently ongoing (start date in the past, end date in the future) should be selected instead of the next enrollable run with a future start date.

How can this be tested?

• Assign XPRO_* env variables in your backend.local.env file
• Run ./manage.py backpopulate_xpro_data
• Wait for the search index to finish updating.
• Search for "Leading Teams" - https://open.odl.local:8063/search?q=%22Leading+Teams%22
• The result card for that course should show "Starts <today's date>", and not "February 9, 2026" (which is the date shown on RC and prod, and locally if you run the above command on the main branch instead)
• You can repeat this for other courses mentioned in the issue.

Screenshots

This branch:

Main branch, RC/Prod:

[ChristopherChudzicki]

👍

[ChristopherChudzicki]

Should this be best_run? It seems to start w/ current run, but fallback to future or older.

[mbertrand]

Will rename to best_lr_run to avoid any confusion with the property name itself.

[ChristopherChudzicki]

I have about 3k resources locally from ocw, xpro, mitxonline, see, and mit_edx. (I think thats all of them?) All published resources have a best_run 👍, but some unpublished ones don't.

If somehow a published resource had no best run, we get errors below, right? Should we keep the best_run check here? What happens if errors below occur? Just that 1 resource fails ETL, right?

[mbertrand]

That should only happen if a resource has no published runs at all, in which case the resource itself shouldn't be published. I'll add the best_run check back just in case.

[ChristopherChudzicki]

👍 Thanks for the explanation!