Adds .gitea/workflows/test-backoffice.yml (paths-filtered to
backoffice/app/** and backoffice/db/**, self-hosted homelab runner,
plain git clone matching deploy-backoffice.yml's conventions) so a
regression no longer needs a human to notice it.
The psycopg[binary]==3.2.1 pin turned out to be correct, not stale: it
only ships wheels through cp312, so it resolves cleanly under the
project's python:3.12-slim target but not under newer interpreters.
Documented in requirements.txt and docs/agents/testing.md, along with
the fact that the suite spins up its own throwaway postgres:16-alpine
container (tests/conftest.py) with no manual DB setup required.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Adds the plumbing that makes "can a customer actually get booked" true
end to end at the API layer, on top of #15's schema/tenancy layer.
- resource_hours table + min_notice_minutes/max_advance_days/buffer_minutes
on resources -- config #15 didn't include but #16 depends on.
- availability.py: pure slot-generation function, correct across a
Europe/Berlin DST transition (tested both directions).
- booking_api.py: JSON blueprint for slot listing, booking creation
(auto_confirm -> confirmed/pending), and signed-JWT cancel/reschedule,
registered into app.py.
- booking_db.py gains resource-hours CRUD, a tenant-scoped busy-bookings
query for buffer/slot validation, and a read-only client lookup.
A true concurrent-threads test (not just sequential requests) surfaced a
real gap: Postgres can raise DeadlockDetected instead of ExclusionViolation
when two overlapping inserts race the exclusion constraint directly, which
went uncaught and would have 500'd instead of giving the clean 4xx the
ticket requires -- now caught alongside ExclusionViolation.
Also fixed: reschedule used the request's raw UTC offset to pick the
business day instead of the client's own timezone (could pick the wrong
day's hours/bookings near local midnight); the cancel/reschedule JWT no
longer falls back to reusing CRM_API_TOKEN as its signing secret.
Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Stand up the DB-first CRM backbone (architecture pivot: Postgres is the
source of truth, Google Sheets becomes a one-way downstream mirror).
- backoffice/ stack: smb-db (Postgres 16) + smb-crm (Flask/waitress service).
- Schema mirrors the six Sheet tabs (clients, leads, projects, activity_log,
bookings, invoices) with typed columns + updated_at triggers.
- Service-account Sheets client (PyJWT) for the one-time import + future mirror.
- import_from_sheets.py: idempotent seed of Postgres from the live Sheets.
- Read dashboard (Leads & Clients tables) at onboard.mivanchenko.de/crm,
behind the existing Caddy basic-auth; JSON API reads straight from Postgres.
Deployed + verified: import seeded DB, dashboard/API live, no-auth blocked,
onboarding form unaffected. Add/edit/delete + DB->Sheets sync + n8n ingest
swap are the next steps.
Co-Authored-By: Claude Opus 4.8 <noreply@anthropic.com>
mivanchenko