feat(website): python seeder with pango lineages and test suite

[fhennig]

Summary

• Replaces the JS seed.mjs with a unified Python seeder in collection-seeding/ (renamed from example-data/)
• Modular source architecture — new data sources are subclasses of the Source ABC; registering a new source is a one-liner in sources/registry.py
• Three sources implemented:
  • covid-resistance-mutations — port of the original JS resistance mutation data (3CLpro, RdRp, Spike mAb)
  • covid-pango-lineages — fetches ~4,976 pango lineage definitions from corneliusroemer/pango-sequences, one collection per lineage with nucleotide substitutions as variants
  • covid-pango-lineages-sample — same as above but limited to 10 lineages, for quick testing (excluded from default run)
• CLI uses --source NAME to run a single source; no flag runs all. --list prints available sources
• Upserts collections (create or update by name) via POST /collections and PUT /collections/{id} — no diffing: every existing collection is always updated on each run
• Deletes orphaned collections — each source declares an owned_tag (e.g. #pango-lineage) which is appended to every collection description. On each run, any existing collection with that tag that is no longer in the source output is deleted via DELETE /collections/{id}
• Authenticates via --api-key / $API_KEY; the system user API key is hardcoded as a dummy value in Docker Compose (both backend and seeder)
• Uses pixi for dependency management with a multi-stage Docker build (pixi builder → python:3.13-slim)
• Typed throughout with TypedDict (Collection, Variant, FilterObject, ExistingCollection)
• Supports repeat scheduling via --repeat-interval-hours CLI flag or $REPEAT_INTERVAL_HOURS env var — if set, the seeder loops with a sleep between runs rather than exiting; Docker Compose runs it every 8 hours with restart: always
• 32 tests covering HTTP interactions, mutation name math, collection building, upsert orchestration, and orphan deletion

Test plan

• pixi run -e test test — all 32 tests pass
• pixi run seed — seeds all sources against a local backend
• pixi run seed again — all collections updated (upsert)
• pixi run seed-lineages — seeds all pango lineages
• pixi run seed-lineages-sample — seeds 10 lineages quickly

[vercel]

The latest updates on your projects. Learn more about Vercel for GitHub.

Project	Deployment	Actions	Updated (UTC)
dashboards	Ready	Preview, Comment	May 29, 2026 1:23pm

[Copilot]

Pull request overview

Copilot reviewed 22 out of 25 changed files in this pull request and generated 3 comments.

[fengelniederhammer]

👍