Books Page: Get Best Edition → Solr

[mekarpeles]

Related to #7451, performance

Describe the problem that you'd like solved

When a reader visits a Book Page, many of its editions are fetched, e.g. for the Other Editions table or to find the best edition: (https://github.com/internetarchive/openlibrary/blob/master/openlibrary/templates/type/edition/view.html#L41-L71). This process requires we fetch all the editions from infogami (rather than solr). This function should make a single solr call (instead of infogami db) -- potentially the same call which is required for loading the #7451. By using solr, we can also do a better job about choosing the desired edition (since we know more about the target language, etc)

Unknowns / Risks

• Solr fields may be different form infobase for editions and we'll want to be careful that the edition we fetch to render the books page is in the format we need -- this may entails functions that are called on the edition model (which solr records won't have)
  • We likely want to use solr to fetch the correct set of representative works (i.e. all books that have some read/preview option -- not limiting to 10 like we are for infobase)
  • Specifically for the edition we're using the load the book page, we may want to fetch this from infobase so it is hydrated with everything we need.

Details

• The books page will fetch…
  • the work from infogami
  • the editions from solr → new logic to select the best one (consider providers too)
  • the edition from infogami
• The goal is eventually to replace the editions table with the editions from solr and also see if we can eliminate the extra call for the edition from infogami (in that order of priority – i.e. editions table, first). But this can all be done later because it’s significantly more work and the clear win is moving editions from infogami to solr.

Needs

A fully complete book record for testing

[mekarpeles]

@cdrini to write docs on which solr work query to use

[cdrini]

The query we'll want is something like:

run_solr_query(
    WorkSearchScheme(),
    f'key:"{work_key}"',
    rows=1,
    fields=['key', 'editions'],
)

That should return a result like this: https://testing.openlibrary.org/search.json?q=key%3A%2Fworks%2FOL24156W&mode=everything&fields=key,editions

Note: I've fixed a bug with this endpoint that was preventing it from working correctly with key , so the link above is for testing. The PR for that bug fix is #8642 .

[jimchamp]

Thanks @cdrini!