Files
GIA/core/management/commands/recalculate_contact_availability.py

252 lines
9.7 KiB
Python

from __future__ import annotations
from django.core.management.base import BaseCommand
from core.models import ContactAvailabilityEvent, ContactAvailabilitySpan, Message
from core.presence import AvailabilitySignal, record_native_signal
from core.presence.inference import now_ms
_SOURCE_ORDER = {
"message_in": 10,
"message_out": 20,
"read_receipt": 30,
"native_presence": 40,
}
class Command(BaseCommand):
help = (
"Recalculate contact availability events/spans from persisted message, "
"read-receipt, and reaction history (deterministic rebuild)."
)
def add_arguments(self, parser):
parser.add_argument("--days", type=int, default=90)
parser.add_argument("--limit", type=int, default=20000)
parser.add_argument("--service", default="")
parser.add_argument("--user-id", default="")
parser.add_argument("--dry-run", action="store_true", default=False)
parser.add_argument("--no-reset", action="store_true", default=False)
def _iter_messages(self, *, days: int, limit: int, service: str, user_id: str):
cutoff_ts = now_ms() - (max(1, int(days)) * 24 * 60 * 60 * 1000)
qs = Message.objects.filter(ts__gte=cutoff_ts).select_related(
"user", "session", "session__identifier", "session__identifier__person"
)
if service:
qs = qs.filter(source_service=str(service).strip().lower())
if user_id:
qs = qs.filter(user_id=str(user_id).strip())
return qs.order_by("ts")[: max(1, int(limit))]
def _build_event_rows(self, messages):
rows = []
for msg in messages:
identifier = getattr(getattr(msg, "session", None), "identifier", None)
person = getattr(identifier, "person", None)
user = getattr(msg, "user", None)
if not identifier or not person or not user:
continue
service = (
str(
getattr(msg, "source_service", "")
or getattr(identifier, "service", "")
)
.strip()
.lower()
)
if not service:
continue
ts = int(getattr(msg, "ts", 0) or 0)
if ts > 0:
author = str(getattr(msg, "custom_author", "") or "").strip().upper()
outgoing = author in {"USER", "BOT"}
rows.append(
{
"user": user,
"person": person,
"person_identifier": identifier,
"service": service,
"source_kind": "message_out" if outgoing else "message_in",
"availability_state": "available",
"confidence": 0.65 if outgoing else 0.75,
"ts": ts,
"payload": {
"origin": "recalculate_contact_availability",
"message_id": str(msg.id),
"inferred_from": "message_activity",
},
}
)
read_ts = int(getattr(msg, "read_ts", 0) or 0)
if read_ts > 0:
rows.append(
{
"user": user,
"person": person,
"person_identifier": identifier,
"service": service,
"source_kind": "read_receipt",
"availability_state": "available",
"confidence": 0.95,
"ts": read_ts,
"payload": {
"origin": "recalculate_contact_availability",
"message_id": str(msg.id),
"inferred_from": "read_receipt",
"read_by": str(
getattr(msg, "read_by_identifier", "") or ""
),
},
}
)
reactions = list(
(getattr(msg, "receipt_payload", {}) or {}).get("reactions") or []
)
for reaction in reactions:
item = dict(reaction or {})
if bool(item.get("removed")):
continue
reaction_ts = int(item.get("updated_at") or 0)
if reaction_ts <= 0:
continue
rows.append(
{
"user": user,
"person": person,
"person_identifier": identifier,
"service": service,
"source_kind": "native_presence",
"availability_state": "available",
"confidence": 0.9,
"ts": reaction_ts,
"payload": {
"origin": "recalculate_contact_availability",
"message_id": str(msg.id),
"inferred_from": "reaction",
"emoji": str(item.get("emoji") or ""),
"actor": str(item.get("actor") or ""),
"source_service": str(
item.get("source_service") or service
),
},
}
)
rows.sort(
key=lambda row: (
str(getattr(row["user"], "id", "")),
str(getattr(row["person"], "id", "")),
str(row.get("service") or ""),
int(row.get("ts") or 0),
_SOURCE_ORDER.get(str(row.get("source_kind") or ""), 999),
str((row.get("payload") or {}).get("message_id") or ""),
)
)
return rows
def handle(self, *args, **options):
days = max(1, int(options.get("days") or 90))
limit = max(1, int(options.get("limit") or 20000))
service_filter = str(options.get("service") or "").strip().lower()
user_filter = str(options.get("user_id") or "").strip()
dry_run = bool(options.get("dry_run"))
reset = not bool(options.get("no_reset"))
cutoff_ts = now_ms() - (days * 24 * 60 * 60 * 1000)
messages = list(
self._iter_messages(
days=days,
limit=limit,
service=service_filter,
user_id=user_filter,
)
)
rows = self._build_event_rows(messages)
keys_to_reset = set()
for row in rows:
keys_to_reset.add(
(
str(getattr(row["user"], "id", "")),
str(getattr(row["person"], "id", "")),
str(row.get("service") or ""),
)
)
deleted_events = 0
deleted_spans = 0
if reset and keys_to_reset and not dry_run:
for user_id, person_id, service in keys_to_reset:
deleted_events += ContactAvailabilityEvent.objects.filter(
user_id=user_id,
person_id=person_id,
service=service,
ts__gte=cutoff_ts,
).delete()[0]
deleted_spans += ContactAvailabilitySpan.objects.filter(
user_id=user_id,
person_id=person_id,
service=service,
end_ts__gte=cutoff_ts,
).delete()[0]
created = 0
dedup_seen = set()
for row in rows:
dedup_key = (
str(getattr(row["user"], "id", "")),
str(getattr(row["person"], "id", "")),
str(getattr(row["person_identifier"], "id", "")),
str(row.get("service") or ""),
str(row.get("source_kind") or ""),
int(row.get("ts") or 0),
str((row.get("payload") or {}).get("message_id") or ""),
str((row.get("payload") or {}).get("inferred_from") or ""),
)
if dedup_key in dedup_seen:
continue
dedup_seen.add(dedup_key)
if not reset:
exists = ContactAvailabilityEvent.objects.filter(
user=row["user"],
person=row["person"],
person_identifier=row["person_identifier"],
service=row["service"],
source_kind=row["source_kind"],
ts=row["ts"],
).exists()
if exists:
continue
created += 1
if dry_run:
continue
record_native_signal(
AvailabilitySignal(
user=row["user"],
person=row["person"],
person_identifier=row["person_identifier"],
service=row["service"],
source_kind=row["source_kind"],
availability_state=row["availability_state"],
confidence=float(row["confidence"]),
ts=int(row["ts"]),
payload=dict(row["payload"]),
)
)
self.stdout.write(
self.style.SUCCESS(
"recalculate_contact_availability complete "
f"messages_scanned={len(messages)} candidates={len(rows)} "
f"created={created} deleted_events={deleted_events} deleted_spans={deleted_spans} "
f"reset={reset} dry_run={dry_run} days={days} limit={limit}"
)
)