"""Phase 4.5 backfill — populate the cross-module participant timeline. Walks the legacy data we already track per-domain and mirrors it into the generic :class:`Participant` / :class:`TrackingRecord` / :class:`TrackingMilestone` schema. Idempotent — safe to re-run because every insert uses ``get_or_create`` keyed on the natural identity of the upstream row. Sources read: * ``SMEHubTrackingRecord`` rows — entrepreneurs + sp1..sp5 milestone buckets. * ``UpstreamEvent`` rows that carry a ``subject`` (User) — mirror to the corresponding source_module timeline. * ``alumni_profiles.AlumniProfile`` rows — link Participants to alumni profile. """ from __future__ import annotations from django.db import migrations from django.utils import timezone _MODULE_BY_UPSTREAM = { "rims": "rims", "rep": "rep", "repository": "repository", "smehub": "smehub", "alumni": "alumni", } def _ensure_participant(Participant, user_id, *, personas, primary): participant, created = Participant.objects.get_or_create( user_id=user_id, defaults={"personas": list(personas), "primary_persona": primary}, ) changed = False existing = list(participant.personas or []) for p in personas: if p not in existing: existing.append(p) changed = True if changed: participant.personas = existing if not participant.primary_persona and primary: participant.primary_persona = primary changed = True if changed: participant.save(update_fields=["personas", "primary_persona", "updated_at"]) return participant def _ensure_record(TrackingRecord, *, participant, source_module): record, _ = TrackingRecord.objects.get_or_create( participant=participant, source_module=source_module, ) return record def _create_milestone(TrackingMilestone, *, record, event_type, source_id, occurred_at, payload, label, source_module): obj, created = TrackingMilestone.objects.get_or_create( record=record, event_type=event_type, source_id=str(source_id), defaults={ "occurred_at": occurred_at, "source_module": source_module, "payload": payload or {}, "label": label, }, ) return obj, created def backfill(apps, schema_editor): Participant = apps.get_model("mel_tracking", "Participant") TrackingRecord = apps.get_model("mel_tracking", "TrackingRecord") TrackingMilestone = apps.get_model("mel_tracking", "TrackingMilestone") SMEHubTrackingRecord = apps.get_model("mel_tracking", "SMEHubTrackingRecord") UpstreamEvent = apps.get_model("mel_tracking", "UpstreamEvent") try: AlumniProfile = apps.get_model("alumni_profiles", "AlumniProfile") except LookupError: AlumniProfile = None # ---- SME-Hub ---- sme_records = ( SMEHubTrackingRecord.objects.all() .select_related("entrepreneur", "business", "baseline_snapshot") ) bucket_to_module = "smehub" for sme in sme_records.iterator(): ent = sme.entrepreneur if ent is None or ent.user_id is None: continue participant = _ensure_participant( Participant, ent.user_id, personas=["entrepreneur"], primary="entrepreneur", ) if participant.entrepreneur_profile_id != ent.pk: participant.entrepreneur_profile_id = ent.pk participant.save(update_fields=["entrepreneur_profile", "updated_at"]) record = _ensure_record( TrackingRecord, participant=participant, source_module=bucket_to_module, ) # Baseline event. if sme.sp1_baseline_at: _create_milestone( TrackingMilestone, record=record, event_type="smehub_baseline_initialised", source_id=f"baseline:{sme.pk}", occurred_at=sme.sp1_baseline_at, payload={"business_id": sme.business_id, "baseline_id": sme.baseline_snapshot_id}, label="SME-Hub baseline initialised", source_module=bucket_to_module, ) # SP2..SP5 buckets. for bucket in ("sp2_milestones", "sp3_partnerships", "sp4_commercialisation", "sp5_investment"): for entry in (getattr(sme, bucket, None) or []): key = entry.get("event") or "milestone" payload = entry.get("payload") or {} ts_iso = entry.get("ts") try: occurred = timezone.datetime.fromisoformat(ts_iso) if ts_iso else timezone.now() except (TypeError, ValueError): occurred = timezone.now() source_id = f"{bucket}:{key}:{payload.get('id') or ts_iso}" _create_milestone( TrackingMilestone, record=record, event_type=f"smehub_{key}", source_id=source_id, occurred_at=occurred, payload={"bucket": bucket, **payload}, label=key.replace("_", " ").title(), source_module=bucket_to_module, ) # Refresh last_event_at on the record. latest = ( TrackingMilestone.objects.filter(record=record) .order_by("-occurred_at") .values_list("occurred_at", flat=True) .first() ) if latest and (record.last_event_at is None or latest > record.last_event_at): record.last_event_at = latest record.save(update_fields=["last_event_at", "updated_at"]) # ---- UpstreamEvent (RIMS / REP / Repository / Alumni when subject present) ---- for ue in UpstreamEvent.objects.exclude(subject__isnull=True).iterator(): module = _MODULE_BY_UPSTREAM.get(ue.module, ue.module) if module not in {"rims", "rep", "repository", "smehub", "alumni"}: continue persona_map = { "rims": "scholar", "rep": "learner", "repository": "researcher", "smehub": "entrepreneur", "alumni": "alumni", } primary = persona_map.get(module) participant = _ensure_participant( Participant, ue.subject_id, personas=[primary] if primary else [], primary=primary or "", ) record = _ensure_record( TrackingRecord, participant=participant, source_module=module, ) _create_milestone( TrackingMilestone, record=record, event_type=ue.event_type, source_id=ue.source_object_id, occurred_at=ue.recorded_at, payload=ue.payload or {}, label=ue.event_type.replace("_", " ").title(), source_module=module, ) latest = ue.recorded_at if record.last_event_at is None or latest > record.last_event_at: record.last_event_at = latest record.save(update_fields=["last_event_at", "updated_at"]) # ---- Link Alumni profiles to participants. ---- if AlumniProfile is not None: for ap in AlumniProfile.objects.iterator(): if ap.user_id is None: continue participant = _ensure_participant( Participant, ap.user_id, personas=["alumni"], primary="alumni", ) if participant.alumni_profile_id != ap.pk: participant.alumni_profile_id = ap.pk participant.save(update_fields=["alumni_profile", "updated_at"]) def reverse_noop(apps, schema_editor): """No-op reverse — Participant/TrackingRecord/TrackingMilestone tables are re-created from forward signals; we don't tear down rebuilt rows.""" return None class Migration(migrations.Migration): dependencies = [ ("mel_tracking", "0006_participant_impactscore_baselinerecord_and_more"), ("alumni_profiles", "0001_initial"), ("smehub_onboarding", "0001_initial"), ] operations = [ migrations.RunPython(backfill, reverse_noop), ]