Import analysis + start import button

[bradenmacdonald]

On both the backend (I think it's missing) and the frontend, follow #2524 by implementing:

• The Course Import Analyzer API that returns the required data (# of each type of block, list of incompatible blocks, etc.)
• The UI that reports on the importability of the course:
  
• The button to start the actual import, if the status is not "blocked".

Note: for now, we always import all content. The option to only import some subset of the course content may come later.

[navinkarkera]

@bradenmacdonald We can fetch all details required in the summary using meilisearch API. For example:

Using below query:

{"queries":[{"q":"","indexUid":"tutor_studio_content","facets":["block_type", "content.problem_types"],"filter":["context_key = \"course-v1:UNIX+UX1+2025_T3\""],"limit":0}]}

We get something like this:

{
  "results": [
    {
      "indexUid": "tutor_studio_content",
      "hits": [],
      "query": "",
      "processingTimeMs": 0,
      "limit": 0,
      "offset": 0,
      "estimatedTotalHits": 124,
      "facetDistribution": {
        "block_type": {
          "chapter": 3,
          "html": 26,
          "itembank": 4,
          "library_content": 8,
          "problem": 42,
          "sequential": 14,
          "vertical": 24,
          "video": 3
        },
        "content.problem_types": {
          "choiceresponse": 9,
          "multiplechoiceresponse": 23,
          "numericalresponse": 4,
          "optionresponse": 4,
          "stringresponse": 1
        }
      },
      "facetStats": {}
    }
  ]
}

• The index however is not always up to date due (even in above example, I only have 2 sections/chapters)
• Instances without meilisearch deployment won't be able to see this info

If the second point is not an issue, we can spend sometime to fix index sync issues and optionally add an option to reindex the course in this page to make sure the data is up to date

[bradenmacdonald]

@navinkarkera Sounds good, I think we should keep it simple and just use Meilisearch then.

The meilisearch requirement is not an issue, because the Libraries v2 UI requires Meilisearch. We should avoid depending on Meilisearch for the actual core migration commands and python API, but it's fine to use it to provide data for this sort of informational summary UI.

The index however is not always up to date due (even in above example, I only have 2 sections/chapters)

Hmm, do you know why?

[navinkarkera]

@bradenmacdonald One known issue is how delete signal is handled and only the deleted parent is removed from index while its children blocks are not deleted

[navinkarkera]

@bradenmacdonald Found a simple way to delete children index data, see this openedx/openedx-platform#37427 (comment)

[navinkarkera]

@sdaitzman @edschema Should we list the type of unsupported blocks here:

[navinkarkera]

@bradenmacdonald @kdmccormick I am listing answers to above questions as per my understanding, let me know if you have any thoughts.

Roughly how long will analysis take for typical courses and for large courses?

Since we are using meilisearch index, should be instant in most cases.

Pre-import are there any other reasons a block would be flagged besides lack of support?

I don't think so.

Confirm: there aren’t any expected cases where structural blocks would be flagged pre-import?

Can't think of any.

Are there any reasons a course would be flagged as blocked pre-import?

Same here, @kdmccormick What do you think?

cc: @sdaitzman @edschema

[navinkarkera]

Also, on clicking Import Course, should we immediately redirect user to Import details page or Import home page?

If the Import details page is shown, it won't have any information so maybe we need to show some kind of inprogress state?

cc: @ChrisChV

[navinkarkera]

@sdaitzman @edschema @kdmccormick @bradenmacdonald Gentle reminder.