[zaira-bibi]

What

Analytics read code was spread over reads.py, a schemas/reads.py that mixed models and enums, and a
top-level enums.py, and its bucket SQL was private even though Explore needs it. This moves it into a
reads package of small, single-purpose files, with no behaviour change.

Changes

• refactor(core): sparkth/core/analytics/reads/ replaces reads.py, schemas/reads.py and enums.py:
  • buckets.py: the PostgreSQL and SQLite bucket SQL and the day-range conditions, now public
    (PG_BUCKET, SQLITE_BUCKET, in_days)
  • rollups.py: read_rollup and its SQL-identifier checks
  • queries.py: get_logins, the core Overview figures and summarize
  • schemas.py: RangeQuery, LoginsPoint, SummaryCount
  • enums.py: Bucket
• refactor(core): the event catalog becomes a package, with its models in catalog/schemas.py and
  FieldKind in catalog/enums.py
• refactor(core): schemas/ now holds only the event schemas
• sparkth.lib.analytics exports the same names, so plugins and routes are unchanged
• chore: the es/fr catalogs point the Overview figure labels at their new file

How to Test

1. uv run pytest -q: the whole suite passes unchanged.
2. make test.frontend.api: the OpenAPI schema and generated client are unchanged.
3. Open Analytics: Overview, Logins and Course authoring show the same figures as before.

Environment Variables

Notes

• Stacked on the event catalog PR; the Explore query PR rebases onto this one.
• No migrations, env vars or dependency changes.

This description was written with the assistance of an LLM (Claude).

🤖 Generated with Claude Code