Files
planet/backend/tests/test_earth_layer_cache.py
2026-11-09 15:12:50 +08:00

558 lines
17 KiB
Python

from __future__ import annotations
import asyncio
import json
import httpx
import pytest
from fastapi import FastAPI, HTTPException, Response
from app.api.v1 import visualization
from app.services.earth_layer_cache import (
EarthLayerCachePolicy,
apply_payload_budget,
earth_layer_cache,
format_bbox_key,
quantize_bbox,
resolve_layer_payload,
)
class FakeRedis:
def __init__(self, *, fail: bool = False) -> None:
self.store: dict[str, str] = {}
self.fail = fail
self.lock_claimed = False
def _maybe_fail(self) -> None:
if self.fail:
raise RuntimeError("redis unavailable")
def get(self, key: str):
self._maybe_fail()
return self.store.get(key)
def set(self, key: str, value: str, nx: bool = False, ex: int | None = None):
self._maybe_fail()
if nx and key in self.store:
return False
self.store[key] = value
return True
def setex(self, key: str, _seconds: int, value: str):
self._maybe_fail()
self.store[key] = value
return True
def delete(self, *keys: str):
self._maybe_fail()
deleted = 0
for key in keys:
deleted += 1 if self.store.pop(key, None) is not None else 0
return deleted
def eval(self, script: str, numkeys: int, *args):
self._maybe_fail()
keys, values = args[:numkeys], args[numkeys:]
if self.store.get(keys[0]) != values[0]:
return 0
if numkeys == 3:
self.setex(keys[1], values[1], values[3])
self.setex(keys[2], values[2], values[3])
return 1
if "'expire'" in script:
return 1
return self.delete(keys[0])
def scan_iter(self, match: str):
self._maybe_fail()
prefix = match.rstrip("*")
for key in list(self.store):
if key.startswith(prefix):
yield key
def memory_usage(self, key: str):
value = self.store.get(key, "")
return len(value.encode("utf-8"))
@pytest.fixture(autouse=True)
def fake_cache_client():
previous = earth_layer_cache._client
fake = FakeRedis()
earth_layer_cache._client = fake
try:
yield fake
finally:
earth_layer_cache._client = previous
def test_quantized_bbox_key_is_stable_for_small_movements():
first = format_bbox_key(quantize_bbox((10.01, 59.04, 10.96, 60.02)))
second = format_bbox_key(quantize_bbox((10.04, 59.01, 10.99, 60.04)))
assert first == second
assert first == "10.0,59.0,11.0,60.0"
def test_payload_budget_truncates_features():
payload = {
"type": "FeatureCollection",
"features": [{"id": index} for index in range(5)],
}
policy = EarthLayerCachePolicy(60, 120, max_features=2, max_bytes=1024)
result = apply_payload_budget(payload, policy)
assert len(result["features"]) == 2
assert result["diagnostics"]["truncated"] is True
assert result["diagnostics"]["limit_reason"] == "feature_budget"
assert result["diagnostics"]["original_feature_count"] == 5
@pytest.mark.asyncio
async def test_resolve_layer_payload_writes_fresh_and_stale(fake_cache_client):
calls = 0
async def builder():
nonlocal calls
calls += 1
return {"type": "FeatureCollection", "features": [{"id": "a"}]}
key = earth_layer_cache.key("satellites", limit="all")
policy = EarthLayerCachePolicy(60, 120)
first = await resolve_layer_payload(key=key, policy=policy, builder=builder)
second = await resolve_layer_payload(key=key, policy=policy, builder=builder)
assert first.state == "refresh"
assert second.state == "hit"
assert calls == 1
assert key in fake_cache_client.store
assert f"{key}:stale" in fake_cache_client.store
@pytest.mark.asyncio
async def test_resolve_layer_payload_returns_stale_when_builder_fails(fake_cache_client):
key = earth_layer_cache.key("bgp-incidents", status="active")
fake_cache_client.store[f"{key}:stale"] = json.dumps(
{"type": "FeatureCollection", "features": []}
)
async def builder():
raise RuntimeError("db exploded")
result = await resolve_layer_payload(
key=key,
policy=EarthLayerCachePolicy(60, 120),
builder=builder,
)
assert result.state == "stale"
assert result.payload["features"] == []
@pytest.mark.asyncio
async def test_resolve_layer_payload_uses_stale_during_lock_contention(fake_cache_client):
key = earth_layer_cache.key("cables")
fake_cache_client.store[earth_layer_cache.lock_key(key)] = "1"
fake_cache_client.store[f"{key}:stale"] = json.dumps(
{"type": "FeatureCollection", "features": [{"id": "stale-cable"}]}
)
calls = 0
async def builder():
nonlocal calls
calls += 1
return {"type": "FeatureCollection", "features": [{"id": "fresh-cable"}]}
result = await resolve_layer_payload(
key=key,
policy=EarthLayerCachePolicy(60, 120),
builder=builder,
)
assert result.state == "stale"
assert result.payload["features"][0]["id"] == "stale-cable"
assert calls == 0
@pytest.mark.asyncio
async def test_resolve_layer_payload_bypasses_redis_failure():
previous = earth_layer_cache._client
earth_layer_cache._client = FakeRedis(fail=True)
try:
async def builder():
return {"type": "FeatureCollection", "features": [{"id": "safe"}]}
result = await resolve_layer_payload(
key=earth_layer_cache.key("cables"),
policy=EarthLayerCachePolicy(60, 120),
builder=builder,
)
assert result.state == "bypass"
assert result.payload["features"][0]["id"] == "safe"
finally:
earth_layer_cache._client = previous
@pytest.mark.asyncio
async def test_visualization_endpoint_sets_cache_headers(fake_cache_client, monkeypatch):
calls = 0
async def fake_build_satellites_geojson(*, limit, db):
nonlocal calls
calls += 1
return {"type": "FeatureCollection", "features": [{"id": f"sat-{limit}"}], "count": 1}
monkeypatch.setattr(visualization, "_build_satellites_geojson", fake_build_satellites_geojson)
first_response = Response()
first = await visualization.get_satellites_geojson(
limit=25, db=object(), response=first_response
)
second_response = Response()
second = await visualization.get_satellites_geojson(
limit=25, db=object(), response=second_response
)
assert first == second
assert calls == 1
assert first_response.headers["X-Planet-Cache"] == "refresh"
assert second_response.headers["X-Planet-Cache"] == "hit"
@pytest.mark.asyncio
async def test_vessel_snapshot_uses_short_cache(fake_cache_client, monkeypatch):
calls = 0
async def fake_load_raw_vessel_snapshot_features(db, *, bbox, limit, observed_since):
nonlocal calls
calls += 1
return (
[
{
"type": "Feature",
"geometry": {"type": "Point", "coordinates": [10.1, 59.1]},
"properties": {"mmsi": 123, "vessel_type_name": "Cargo"},
}
],
{"raw_feature_count": 1},
)
monkeypatch.setattr(
visualization,
"_load_raw_vessel_snapshot_features",
fake_load_raw_vessel_snapshot_features,
)
first_response = Response()
first = await visualization.build_vessel_snapshot_response(
object(),
bbox=(10.01, 59.04, 10.96, 60.02),
zoom=12,
type_filter=None,
limit=1000,
since_minutes=60,
response=first_response,
)
second_response = Response()
second = await visualization.build_vessel_snapshot_response(
object(),
bbox=(10.04, 59.01, 10.99, 60.04),
zoom=12,
type_filter=None,
limit=1000,
since_minutes=60,
response=second_response,
)
assert first["count"] == 1
assert second == first
assert calls == 1
assert first_response.headers["X-Planet-Cache"] == "refresh"
assert second_response.headers["X-Planet-Cache"] == "hit"
@pytest.mark.asyncio
async def test_concurrent_summary_requests_share_one_slow_build(monkeypatch):
started = asyncio.Event()
finish = asyncio.Event()
calls = 0
async def build_summary(db):
nonlocal calls
calls += 1
started.set()
await finish.wait()
return {"stats": {"vessel_count": 42}}
monkeypatch.setattr(visualization, "_build_visualization_geo_summary", build_summary)
app = FastAPI()
app.include_router(visualization.router, prefix="/api/v1/visualization")
app.dependency_overrides[visualization.get_db] = lambda: object()
async with httpx.AsyncClient(
transport=httpx.ASGITransport(app=app), base_url="http://test"
) as client:
requests = [
asyncio.create_task(client.get("/api/v1/visualization/geo/summary")) for _ in range(6)
]
await asyncio.wait_for(started.wait(), 1)
# Keep the build open beyond the old 200 ms fallback to duplicate SQL.
await asyncio.sleep(0.25)
assert calls == 1
finish.set()
results = await asyncio.gather(*requests)
assert all(response.status_code == 200 for response in results)
assert all(response.json() == {"stats": {"vessel_count": 42}} for response in results)
assert sorted(response.headers["X-Planet-Cache"] for response in results) == [
"hit",
"hit",
"hit",
"hit",
"hit",
"refresh",
]
assert not earth_layer_cache._inflight
@pytest.mark.asyncio
@pytest.mark.parametrize("error_type", [RuntimeError, TimeoutError])
async def test_shared_builder_error_is_not_retried(error_type):
started = asyncio.Event()
finish = asyncio.Event()
calls = 0
async def builder():
nonlocal calls
calls += 1
started.set()
await finish.wait()
raise error_type("db failed")
requests = [
asyncio.create_task(
resolve_layer_payload(
key="shared-error", policy=EarthLayerCachePolicy(60, 120), builder=builder
)
)
for _ in range(4)
]
await started.wait()
await asyncio.sleep(0)
finish.set()
results = await asyncio.gather(*requests, return_exceptions=True)
assert calls == 1
assert all(isinstance(result, error_type) for result in results)
assert not earth_layer_cache._inflight
@pytest.mark.asyncio
async def test_concurrent_redis_failure_shares_bypass_result(fake_cache_client):
fake_cache_client.fail = True
started = asyncio.Event()
finish = asyncio.Event()
calls = 0
async def builder():
nonlocal calls
calls += 1
started.set()
await finish.wait()
return {"count": 42}
requests = [
asyncio.create_task(
resolve_layer_payload(
key="shared-bypass", policy=EarthLayerCachePolicy(60, 120), builder=builder
)
)
for _ in range(4)
]
await started.wait()
await asyncio.sleep(0)
finish.set()
results = await asyncio.gather(*requests)
assert calls == 1
assert all(result.state == "bypass" and result.payload == {"count": 42} for result in results)
@pytest.mark.asyncio
async def test_cancelled_waiter_does_not_cancel_owner():
started = asyncio.Event()
finish = asyncio.Event()
policy = EarthLayerCachePolicy(60, 120)
async def builder():
started.set()
await finish.wait()
return {"count": 1}
owner = asyncio.create_task(
resolve_layer_payload(key="cancel-waiter", policy=policy, builder=builder)
)
await started.wait()
waiter = asyncio.create_task(
resolve_layer_payload(key="cancel-waiter", policy=policy, builder=builder)
)
await asyncio.sleep(0)
waiter.cancel()
with pytest.raises(asyncio.CancelledError):
await waiter
finish.set()
assert (await owner).payload == {"count": 1}
assert not earth_layer_cache._inflight
@pytest.mark.asyncio
async def test_cancelled_owner_allows_waiter_to_use_its_own_builder():
started = asyncio.Event()
policy = EarthLayerCachePolicy(60, 120)
async def owner_builder():
started.set()
await asyncio.Event().wait()
async def waiter_builder():
return {"count": 2}
owner = asyncio.create_task(
resolve_layer_payload(key="cancel-owner", policy=policy, builder=owner_builder)
)
await started.wait()
waiter = asyncio.create_task(
resolve_layer_payload(key="cancel-owner", policy=policy, builder=waiter_builder)
)
await asyncio.sleep(0)
owner.cancel()
with pytest.raises(asyncio.CancelledError):
await owner
assert (await asyncio.wait_for(waiter, 1)).payload == {"count": 2}
assert not earth_layer_cache._inflight
@pytest.mark.asyncio
async def test_cross_process_lock_waits_for_payload_instead_of_building(fake_cache_client):
key = "remote-owner"
fake_cache_client.store[earth_layer_cache.lock_key(key)] = "another-owner"
policy = EarthLayerCachePolicy(60, 120, lock_wait_seconds=0.005)
calls = 0
async def builder():
nonlocal calls
calls += 1
return {"count": -1}
waiter = asyncio.create_task(resolve_layer_payload(key=key, policy=policy, builder=builder))
await asyncio.sleep(0.03)
assert calls == 0
fake_cache_client.store[key] = json.dumps({"count": 3})
assert (await asyncio.wait_for(waiter, 1)).payload == {"count": 3}
assert calls == 0
@pytest.mark.asyncio
async def test_wait_timeout_does_not_bypass_other_owner(fake_cache_client):
key = "remote-timeout"
fake_cache_client.store[earth_layer_cache.lock_key(key)] = "another-owner"
calls = 0
async def builder():
nonlocal calls
calls += 1
return {}
with pytest.raises(HTTPException) as error:
await resolve_layer_payload(
key=key,
policy=EarthLayerCachePolicy(60, 120, lock_wait_seconds=0.005, build_wait_seconds=0.02),
builder=builder,
)
assert error.value.status_code == 503
assert calls == 0
assert not earth_layer_cache._inflight
@pytest.mark.asyncio
async def test_lost_owner_cannot_publish_or_release_new_lock(fake_cache_client):
key = "lock-replaced"
policy = EarthLayerCachePolicy(60, 120)
async def builder():
fake_cache_client.store[earth_layer_cache.lock_key(key)] = "new-owner"
fake_cache_client.store[key] = json.dumps({"count": 99})
return {"count": 1}
result = await resolve_layer_payload(key=key, policy=policy, builder=builder)
assert result.payload == {"count": 1}
assert result.state == "bypass"
assert fake_cache_client.store[earth_layer_cache.lock_key(key)] == "new-owner"
assert json.loads(fake_cache_client.store[key]) == {"count": 99}
@pytest.mark.asyncio
async def test_redis_publish_failure_returns_built_payload_without_retry(fake_cache_client):
calls = 0
async def builder():
nonlocal calls
calls += 1
fake_cache_client.fail = True
return {"count": 7}
result = await resolve_layer_payload(
key="publish-error", policy=EarthLayerCachePolicy(60, 120), builder=builder
)
assert result.state == "bypass"
assert result.payload == {"count": 7}
assert calls == 1
@pytest.mark.asyncio
async def test_local_wait_timeout_leaves_owner_running():
started = asyncio.Event()
finish = asyncio.Event()
policy = EarthLayerCachePolicy(60, 120, build_wait_seconds=0.02)
async def builder():
started.set()
await finish.wait()
return {"count": 1}
owner = asyncio.create_task(
resolve_layer_payload(key="local-timeout", policy=policy, builder=builder)
)
await started.wait()
with pytest.raises(HTTPException) as error:
await resolve_layer_payload(key="local-timeout", policy=policy, builder=builder)
assert error.value.status_code == 503
assert not owner.done()
finish.set()
assert (await owner).payload == {"count": 1}
assert not earth_layer_cache._inflight
@pytest.mark.asyncio
async def test_different_keys_can_build_concurrently():
started = set()
both_started = asyncio.Event()
async def build(key):
started.add(key)
if len(started) == 2:
both_started.set()
await asyncio.wait_for(both_started.wait(), 1)
return {"key": key}
results = await asyncio.gather(
*(
resolve_layer_payload(
key=key, policy=EarthLayerCachePolicy(60, 120), builder=lambda key=key: build(key)
)
for key in ("one", "two")
)
)
assert {result.payload["key"] for result in results} == {"one", "two"}