Fix embeddings healthcheck missing-summaries filter

[blarghmatey]

What are the relevant tickets?

N/A

Description (What does it do?)

_missing_summaries() in vector_search/tasks.py calls ContentSummarizer.get_unprocessed_content_file_ids(self, overwrite, learning_resource_ids=None, content_file_ids=None), but was passing a LearningResource queryset positionally:

def _missing_summaries():
    summarizer = ContentSummarizer()
    return summarizer.get_unprocessed_content_file_ids(
        LearningResource.objects.filter(require_summaries=True)
        .filter(Q(published=True) | Q(test_mode=True))
        .values_list("id", flat=True)
    )

That queryset landed in the overwrite parameter instead of learning_resource_ids. Since a non-empty queryset is truthy, this silently set overwrite=True, which skips the "only unprocessed" filter (Q(summary="") | Q(flashcards=[])) inside get_unprocessed_content_file_ids. And because learning_resource_ids was never actually passed, the require_summaries=True scoping was lost entirely.

The net effect: the embeddings_healthcheck Sentry alert this feeds (vector_search/tasks.py) was counting every content file across every learning resource — including ones that already have summaries and ones from resources that don't even require summaries — as "missing," rather than just genuinely unprocessed content on resources that require summaries.

Fix: pass both arguments as keywords (overwrite=False, learning_resource_ids=resource_ids). Also added a short-circuit for the case where no learning resources currently require summaries — get_unprocessed_content_file_ids treats an empty learning_resource_ids list the same as None (no restriction) rather than "match nothing," so without the short-circuit that edge case would still fall through to scanning every resource.

This was flagged by Sentry's bug-prediction bot as a review comment on #3683 — it's a pre-existing bug (introduced in commit 3ba27bdc4e, 2025-11-07) unrelated to that PR's changes, so it's being fixed here separately.

Screenshots (if appropriate):

N/A - no UI changes

How can this be tested?

Added two regression tests to vector_search/tasks_test.py:

• test_embeddings_healthcheck_excludes_already_summarized — a content file that already has a summary and flashcards should not be counted as missing.
• test_embeddings_healthcheck_summaries_scoped_to_require_summaries — a content file belonging to a learning resource with require_summaries=False should not be counted as missing.

Both fail against the old code (the first because overwrite ends up truthy, the second because learning_resource_ids is never applied) and pass with the fix.

Ran the full vector_search/tasks_test.py suite: docker compose run --rm web uv run pytest vector_search/tasks_test.py -q — 45 passed, including the existing test_embeddings_healthcheck_missing_summaries test.

ruff check --select F401,F821 and ruff format --diff are clean on both touched files.

Additional Context

No behavior change outside of _missing_summaries() / the embeddings_healthcheck task — this only affects what gets reported to Sentry, not any content-processing task itself.

[github-actions]

OpenAPI Changes

No changes detected

View full changelog

Unexpected changes? Ensure your branch is up-to-date with main (consider rebasing).

[Copilot]

Pull request overview

Fixes a bug in the embeddings_healthcheck path where _missing_summaries() was passing a LearningResource queryset positionally into ContentSummarizer.get_unprocessed_content_file_ids, causing incorrect scoping and effectively treating all content files as “missing summaries” in Sentry.

Changes:

• Pass overwrite=False and learning_resource_ids=... as keyword arguments to get_unprocessed_content_file_ids to avoid positional mis-binding.
• Add a short-circuit in _missing_summaries() to return [] when no learning resources require summaries (preventing an unintended “scan everything” behavior).
• Add regression tests ensuring the healthcheck excludes already-summarized content files and is scoped to require_summaries=True resources.

Reviewed changes

Copilot reviewed 2 out of 2 changed files in this pull request and generated no comments.

File	Description
vector_search/tasks.py	Fixes _missing_summaries() argument binding and adds an empty-scope short-circuit to keep the healthcheck correctly scoped.
vector_search/tasks_test.py	Adds regression coverage for the corrected scoping and overwrite behavior in embeddings_healthcheck.

[shanbady]

👍