Import Action Network events (incl. campaign events) with scheduled refresh

[joaquimds]

What

Adds event import to the Action Network data source adaptor (previously people-only), plus a daily scheduled refresh since Action Network has no events webhook.

Why the two-crawl approach

Action Network events live in two places: the flat /api/v2/events collection and inside event campaigns (/event_campaigns/{id}/events). Some events only appear via a campaign. Verified against a real key:

Set	Events
Flat /events	12
Distinct across campaigns	27
In both	2
Only reachable via campaigns	25
Total union	37

Importing from /events alone would miss ~68% of events. So the adaptor crawls campaigns first (capturing each campaign's title), then the flat collection, deduping by event id. A live run of the adaptor yielded 37 events, all geocoded, 27 with campaign names, zero duplicates.

Changes

• Config: recordType: "people" | "events" on the Action Network config (defaults to people), with a migration backfilling existing sources to people.
• Adaptor: two-crawl + dedupe event fetch; normalizeEvent flattens title/dates/status/venue/address/postcode and Action Network's server-side coordinates; fetchFirst works for campaign-only orgs; events are read-only (no write-back / tagging).
• Create flow: event sources default to Coordinates geocoding on latitude/longitude with title as name and start_date as date column — map-ready with no geocoding pass.
• UI: People / Events selector in the new data source form.
• Scheduled refresh: importActionNetworkEventDataSources sweep job (mirrors importZetkinDataSources), registered in the worker and scheduled at 5am daily.
• Tests: mocked unit tests for the crawl/dedupe/read-only behaviour.

Scope

Read-only import (no attendances/RSVPs, no write-back) and daily cron refresh (no live webhook — Action Network doesn't offer one for events).

Migration

Run npm run migrate to apply 1784819176587_action_network_config_record_type.

🤖 Generated with Claude Code