Port presence feature forward from orphaned agent/dd2a126b57a1 branch

Recovers a real August feature that was never merged (3 WIP checkpoints +
fix commit 47ab492, 225 commits behind main): a Redis-backed heartbeat
presence store plus /api/presence/{heartbeat,presence,status} endpoints.
Ported onto current main's conventions rather than rebuilt:

- app/models/user.py: last_seen_at/status_message columns (unchanged from
  the orphaned branch; still compatible with the current model).
- alembic/versions/0006_add_presence_columns.py: same column migration,
  rebased as the new head after 0005_add_world_content_tables (was
  2a9b9873244a off 0001_initial).
- app/routers/presence.py: same three endpoints, updated to the DbDep
  pattern other routers (avatar.py) now use; retains the 47ab492 fix
  (is_online_batch via MGET, correct StatusUpdate Field annotation).
- app/services/presence.py: real adaptation beyond a mechanical port --
  the original PresenceStore required an explicit connect() that nothing
  ever called (main.py's lifespan never wired it up), so every real
  request would have raised "not connected". Made the Redis client connect
  lazily on first use instead, and cache it per running event loop rather
  than globally: this repo's TestClient(app) pattern (no `with` block) can
  dispatch different requests onto different event loops, which broke a
  single cached connection the same way NullPool's comment in conftest.py
  describes for asyncpg. Dropped the unused get_presence()/
  update_status_message() methods and the never-wired init/shutdown
  singleton helpers -- dead code even in the original.
- tests/conftest.py: extended the existing ephemeral-Postgres bootstrap to
  also bring up ephemeral Redis (docker-compose.yml already defined the
  service) when REDIS_URL isn't already reachable, so `pytest` stays
  hermetic.
- tests/test_presence.py: heartbeat sets online, batch presence over
  mixed online/offline/unknown ids (the exact 47ab492 regression), status
  message update/strip/length-limit, and TTL expiry via a short-TTL
  PresenceStore instance (real Redis SET/EXISTS/MGET, no 60s sleep needed).

Verified: full backend pytest suite (144 tests) passes against a fresh
Postgres+Redis. Also drove the real endpoints end-to-end against a running
uvicorn + docker-compose Postgres/Redis: heartbeat -> online, batch query,
forced TTL expiry -> offline with last_seen_at retained, status message
trim + 422 over 140 chars.

Co-Authored-By: Claude Sonnet 5 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Y2FESCaMXeJpC9gSctiEeD
This commit is contained in:
2026-09-14 23:59:14 +00:00
parent 9f04cef447
commit f34a4806ef
7 changed files with 623 additions and 38 deletions
+71 -36
View File
@@ -1,15 +1,18 @@
"""Shared pytest fixtures + hermetic-DB bootstrap.
"""Shared pytest fixtures + hermetic-DB/Redis bootstrap.
The integration tests hit a real Postgres (Postgres-specific types: JSONB,
native ENUMs). To make ``cd backend && pytest`` work from a clean checkout
with no manual setup, ``pytest_configure`` below brings up an ephemeral
Postgres via docker compose and runs the migrations IF ``DATABASE_URL`` is
not already pointing at a reachable database. When it is (dev loop, or the
checkpoint runner which stands the DB up itself), that path is skipped and
the existing database is reused.
native ENUMs) and, for presence tests, a real Redis (TTL semantics aren't
worth faking). To make ``cd backend && pytest`` work from a clean checkout
with no manual setup, ``pytest_configure`` below brings up ephemeral
Postgres/Redis containers via docker compose and runs the migrations IF
``DATABASE_URL``/``REDIS_URL`` aren't already pointing at something
reachable. When they are (dev loop, or the checkpoint runner which stands
these up itself), that path is skipped and the existing services are
reused.
If Docker isn't available and no DB is reachable, the whole DB-backed suite
is skipped with a clear message rather than erroring out 60+ times.
If Docker isn't available and nothing is reachable, the whole
DB/Redis-backed suite is skipped with a clear message rather than erroring
out dozens of times.
"""
import asyncio
@@ -44,8 +47,9 @@ TEST_TABLES = (
BACKEND_DIR = Path(__file__).resolve().parents[1]
_COMPOSE_PROJECT = "agp-pytest"
_EPHEMERAL_PG_PORT = 25432
_EPHEMERAL_REDIS_PORT = 26379
# Set by _ensure_database() when it starts a container it owns; torn down in
# Set by pytest_configure() when it starts container(s) it owns; torn down in
# pytest_unconfigure.
_started_container = False
_db_unavailable_reason: str | None = None
@@ -64,6 +68,11 @@ def _db_reachable(url: str) -> bool:
return _port_open(parts.hostname or "localhost", parts.port or 5432)
def _redis_reachable(url: str) -> bool:
parts = urlsplit(url)
return _port_open(parts.hostname or "localhost", parts.port or 6379)
def _run(cmd: list[str], **kw) -> subprocess.CompletedProcess:
return subprocess.run(cmd, cwd=BACKEND_DIR, capture_output=True, text=True, **kw)
@@ -71,14 +80,17 @@ def _run(cmd: list[str], **kw) -> subprocess.CompletedProcess:
def pytest_configure(config):
global _started_container, _db_unavailable_reason
if _db_reachable(settings.DATABASE_URL):
return # dev loop / checkpoint runner already provided one
need_db = not _db_reachable(settings.DATABASE_URL)
need_redis = not _redis_reachable(settings.REDIS_URL)
if not need_db and not need_redis:
return # dev loop / checkpoint runner already provided both
if not _run(["docker", "--version"]).returncode == 0:
_db_unavailable_reason = (
"no reachable DATABASE_URL and docker is unavailable -- run the "
"checkpoint suite (scripts/run_checkpoint_tests.py) or start Postgres "
"manually"
"no reachable DATABASE_URL/REDIS_URL and docker is unavailable -- run "
"the checkpoint suite (scripts/run_checkpoint_tests.py) or start "
"Postgres/Redis manually"
)
return
@@ -86,33 +98,55 @@ def pytest_configure(config):
env["COMPOSE_PROJECT_NAME"] = _COMPOSE_PROJECT
env["POSTGRES_HOST_PORT"] = str(_EPHEMERAL_PG_PORT)
env["POSTGRES_PORT"] = str(_EPHEMERAL_PG_PORT)
up = _run(["docker", "compose", "up", "-d", "postgres"], env=env)
env["REDIS_HOST_PORT"] = str(_EPHEMERAL_REDIS_PORT)
services = [s for s, need in (("postgres", need_db), ("redis", need_redis)) if need]
up = _run(["docker", "compose", "up", "-d", *services], env=env)
if up.returncode != 0:
_db_unavailable_reason = f"docker compose up postgres failed:\n{up.stderr[-800:]}"
_db_unavailable_reason = f"docker compose up {' '.join(services)} failed:\n{up.stderr[-800:]}"
return
_started_container = True
for _ in range(30):
r = _run(
["docker", "compose", "exec", "-T", "postgres",
"pg_isready", "-U", "agp_user", "-d", "agp_db"],
env=env,
)
if r.returncode == 0:
break
time.sleep(1)
else:
_db_unavailable_reason = "ephemeral Postgres never became ready"
return
if need_db:
for _ in range(30):
r = _run(
["docker", "compose", "exec", "-T", "postgres",
"pg_isready", "-U", "agp_user", "-d", "agp_db"],
env=env,
)
if r.returncode == 0:
break
time.sleep(1)
else:
_db_unavailable_reason = "ephemeral Postgres never became ready"
return
url = f"postgresql+asyncpg://agp_user:agp_password@localhost:{_EPHEMERAL_PG_PORT}/agp_db"
os.environ["DATABASE_URL"] = url
settings.DATABASE_URL = url
url = f"postgresql+asyncpg://agp_user:agp_password@localhost:{_EPHEMERAL_PG_PORT}/agp_db"
os.environ["DATABASE_URL"] = url
settings.DATABASE_URL = url
mig = _run([sys.executable, "-m", "alembic", "upgrade", "head"],
env={**os.environ, "DATABASE_URL": url})
if mig.returncode != 0:
_db_unavailable_reason = f"alembic upgrade head failed:\n{mig.stderr[-800:]}"
mig = _run([sys.executable, "-m", "alembic", "upgrade", "head"],
env={**os.environ, "DATABASE_URL": url})
if mig.returncode != 0:
_db_unavailable_reason = f"alembic upgrade head failed:\n{mig.stderr[-800:]}"
return
if need_redis:
for _ in range(30):
r = _run(
["docker", "compose", "exec", "-T", "redis", "redis-cli", "ping"],
env=env,
)
if r.returncode == 0 and "PONG" in r.stdout:
break
time.sleep(1)
else:
_db_unavailable_reason = "ephemeral Redis never became ready"
return
redis_url = f"redis://localhost:{_EPHEMERAL_REDIS_PORT}/0"
os.environ["REDIS_URL"] = redis_url
settings.REDIS_URL = redis_url
def pytest_unconfigure(config):
@@ -122,6 +156,7 @@ def pytest_unconfigure(config):
env["COMPOSE_PROJECT_NAME"] = _COMPOSE_PROJECT
env["POSTGRES_HOST_PORT"] = str(_EPHEMERAL_PG_PORT)
env["POSTGRES_PORT"] = str(_EPHEMERAL_PG_PORT)
env["REDIS_HOST_PORT"] = str(_EPHEMERAL_REDIS_PORT)
_run(["docker", "compose", "down", "-v"], env=env)
+221
View File
@@ -0,0 +1,221 @@
"""Integration tests for the presence feature:
- POST /api/presence/heartbeat
- GET /api/presence/presence
- PATCH /api/presence/status
Hits the real app against the configured Postgres *and* a real Redis
(conftest.py brings up an ephemeral instance of each when neither is
already reachable) -- presence is fundamentally "does a TTL key exist in
Redis", which isn't worth faking with a mock.
Also covers the regression fixed in 47ab492 ("Fix StatusUpdate field
annotation and is_online batch-call mismatch"): the router used to call
``PresenceStore.is_online(ids)`` with a *list* of ids and then evaluate
``uid in online_ids`` against the returned bool, which raised a TypeError
on the first real multi-id batch call, and ``StatusUpdate.status_message``
used the invalid annotation ``str | Field(...)``. Both are exercised below.
"""
import asyncio
import pytest
from fastapi.testclient import TestClient
from app.main import app
from app.services.presence import PresenceStore
def _register(client: TestClient, username: str) -> dict:
resp = client.post(
"/api/auth/register",
json={
"username": username,
"email": f"{username}@example.com",
"password": "secret123",
"date_of_birth": "1990-01-01",
},
)
assert resp.status_code == 201, resp.text
token = resp.json()["access_token"]
headers = {"Authorization": f"Bearer {token}"}
me = client.get("/api/users/me", headers=headers)
assert me.status_code == 200, me.text
return {"headers": headers, "id": me.json()["id"]}
class TestHeartbeat:
def test_heartbeat_sets_online(self):
client = TestClient(app)
user = _register(client, "presence_hb1")
resp = client.post("/api/presence/heartbeat", json={}, headers=user["headers"])
assert resp.status_code == 200, resp.text
assert resp.json() == {"status": "ok", "user_id": user["id"]}
presence = client.get(
"/api/presence/presence",
params={"user_ids": str(user["id"])},
headers=user["headers"],
)
assert presence.status_code == 200, presence.text
body = presence.json()["users"]
assert len(body) == 1
assert body[0]["user_id"] == user["id"]
assert body[0]["is_online"] is True
assert body[0]["last_seen_at"] is not None
def test_requires_auth(self):
client = TestClient(app)
resp = client.post("/api/presence/heartbeat", json={})
assert resp.status_code == 401
def test_never_seen_user_is_offline_with_null_last_seen(self):
client = TestClient(app)
user = _register(client, "presence_never_hb")
presence = client.get(
"/api/presence/presence",
params={"user_ids": str(user["id"])},
headers=user["headers"],
)
assert presence.status_code == 200
body = presence.json()["users"][0]
assert body["is_online"] is False
assert body["last_seen_at"] is None
class TestBatchPresence:
"""Covers the batch-call path fixed in 47ab492 (is_online_batch via MGET)."""
def test_batch_query_mixes_online_offline_and_unknown_ids(self):
client = TestClient(app)
online_user = _register(client, "presence_batch_online")
offline_user = _register(client, "presence_batch_offline")
client.post("/api/presence/heartbeat", json={}, headers=online_user["headers"])
unknown_id = max(online_user["id"], offline_user["id"]) + 999_999
ids = f"{online_user['id']},{offline_user['id']},{unknown_id}"
resp = client.get(
"/api/presence/presence",
params={"user_ids": ids},
headers=online_user["headers"],
)
assert resp.status_code == 200, resp.text
by_id = {u["user_id"]: u for u in resp.json()["users"]}
# Unknown id is silently dropped (not in the DB), not a 500 or 404.
assert unknown_id not in by_id
assert by_id[online_user["id"]]["is_online"] is True
assert by_id[offline_user["id"]]["is_online"] is False
def test_empty_user_ids_is_a_400(self):
client = TestClient(app)
user = _register(client, "presence_empty_ids")
resp = client.get(
"/api/presence/presence",
params={"user_ids": ""},
headers=user["headers"],
)
assert resp.status_code == 400
class TestStatusMessage:
def test_update_and_read_back_status_message(self):
client = TestClient(app)
user = _register(client, "presence_status1")
resp = client.patch(
"/api/presence/status",
json={"status_message": " brb, making coffee "},
headers=user["headers"],
)
assert resp.status_code == 200, resp.text
# The router strips whitespace (StatusUpdate.strip_message).
assert resp.json()["status_message"] == "brb, making coffee"
presence = client.get(
"/api/presence/presence",
params={"user_ids": str(user["id"])},
headers=user["headers"],
)
assert presence.json()["users"][0]["status_message"] == "brb, making coffee"
def test_status_message_over_max_length_is_rejected(self):
client = TestClient(app)
user = _register(client, "presence_status_toolong")
resp = client.patch(
"/api/presence/status",
json={"status_message": "x" * 141},
headers=user["headers"],
)
# StatusUpdate.status_message: str | None = Field(..., max_length=140)
# This is exactly the annotation 47ab492 fixed (it used to be the
# invalid `str | Field(...)`, which breaks model construction
# entirely rather than validating length).
assert resp.status_code == 422
def test_status_message_at_max_length_is_accepted(self):
client = TestClient(app)
user = _register(client, "presence_status_maxlen")
resp = client.patch(
"/api/presence/status",
json={"status_message": "x" * 140},
headers=user["headers"],
)
assert resp.status_code == 200, resp.text
def test_null_status_message_clears_it(self):
client = TestClient(app)
user = _register(client, "presence_status_clear")
client.patch("/api/presence/status", json={"status_message": "hi"}, headers=user["headers"])
resp = client.patch("/api/presence/status", json={"status_message": None}, headers=user["headers"])
assert resp.status_code == 200
assert resp.json()["status_message"] is None
class TestPresenceStoreTTL:
"""Exercises the real Redis TTL semantics directly against the service,
with a short heartbeat_ttl so the test doesn't need to sleep for the
real 60s default used in production. This is the cheap, real proxy for
"TTL expiry means offline" the task asked for -- same code path
(real Redis SET ... EX / EXISTS / MGET), just a much shorter window.
"""
async def test_ttl_expiry_flips_online_to_offline(self):
store = PresenceStore(heartbeat_ttl=1)
try:
uid = 987654321 # doesn't need to exist in Postgres; Redis-only check
assert await store.is_online(uid) is False
await store.heartbeat(uid)
assert await store.is_online(uid) is True
await asyncio.sleep(1.5)
assert await store.is_online(uid) is False
finally:
await store.close()
async def test_is_online_batch_matches_is_online_semantics(self):
"""Regression coverage for 47ab492: is_online_batch must return a
set of ids usable with `in`, not a bool."""
store = PresenceStore(heartbeat_ttl=30)
try:
online_uid, offline_uid = 111222333, 444555666
await store.heartbeat(online_uid)
result = await store.is_online_batch([online_uid, offline_uid])
assert isinstance(result, set)
assert online_uid in result
assert offline_uid not in result
finally:
await store.close()
async def test_is_online_batch_empty_list(self):
store = PresenceStore()
try:
assert await store.is_online_batch([]) == set()
finally:
await store.close()