Improve Explore SQL query memory usage (#14451)

* Remove sql window function in explore endpoint

* don't revalidate first page on every fetch
This commit is contained in:
Josh Hawkins 2024-10-19 23:12:54 -05:00 committed by GitHub
parent 2137de37b9
commit b24d292ade
No known key found for this signature in database
GPG Key ID: B5690EEEBB952194
2 changed files with 52 additions and 57 deletions

View File

@ -259,66 +259,61 @@ def events(params: EventsQueryParams = Depends()):
@router.get("/events/explore") @router.get("/events/explore")
def events_explore(limit: int = 10): def events_explore(limit: int = 10):
subquery = Event.select( # get distinct labels for all events
Event.id, distinct_labels = Event.select(Event.label).distinct().order_by(Event.label)
Event.camera,
Event.label,
Event.zones,
Event.start_time,
Event.end_time,
Event.has_clip,
Event.has_snapshot,
Event.plus_id,
Event.retain_indefinitely,
Event.sub_label,
Event.top_score,
Event.false_positive,
Event.box,
Event.data,
fn.rank()
.over(partition_by=[Event.label], order_by=[Event.start_time.desc()])
.alias("rank"),
fn.COUNT(Event.id).over(partition_by=[Event.label]).alias("event_count"),
).alias("subquery")
query = ( label_counts = {}
Event.select(
subquery.c.id,
subquery.c.camera,
subquery.c.label,
subquery.c.zones,
subquery.c.start_time,
subquery.c.end_time,
subquery.c.has_clip,
subquery.c.has_snapshot,
subquery.c.plus_id,
subquery.c.retain_indefinitely,
subquery.c.sub_label,
subquery.c.top_score,
subquery.c.false_positive,
subquery.c.box,
subquery.c.data,
subquery.c.event_count,
)
.from_(subquery)
.where(subquery.c.rank <= limit)
.order_by(subquery.c.event_count.desc(), subquery.c.start_time.desc())
.dicts()
)
events = list(query.iterator()) def event_generator():
for label_obj in distinct_labels.iterator():
label = label_obj.label
processed_events = [ # get most recent events for this label
{k: v for k, v in event.items() if k != "data"} label_events = (
| { Event.select()
"data": { .where(Event.label == label)
k: v .order_by(Event.start_time.desc())
for k, v in event["data"].items() .limit(limit)
if k in ["type", "score", "top_score", "description"] .iterator()
)
# count total events for this label
label_counts[label] = Event.select().where(Event.label == label).count()
yield from label_events
def process_events():
for event in event_generator():
processed_event = {
"id": event.id,
"camera": event.camera,
"label": event.label,
"zones": event.zones,
"start_time": event.start_time,
"end_time": event.end_time,
"has_clip": event.has_clip,
"has_snapshot": event.has_snapshot,
"plus_id": event.plus_id,
"retain_indefinitely": event.retain_indefinitely,
"sub_label": event.sub_label,
"top_score": event.top_score,
"false_positive": event.false_positive,
"box": event.box,
"data": {
k: v
for k, v in event.data.items()
if k in ["type", "score", "top_score", "description"]
},
"event_count": label_counts[event.label],
} }
} yield processed_event
for event in events
] # convert iterator to list and sort
processed_events = sorted(
process_events(),
key=lambda x: (x["event_count"], x["start_time"]),
reverse=True,
)
return JSONResponse(content=processed_events) return JSONResponse(content=processed_events)

View File

@ -177,7 +177,7 @@ export default function Explore() {
const { data, size, setSize, isValidating, mutate } = useSWRInfinite< const { data, size, setSize, isValidating, mutate } = useSWRInfinite<
SearchResult[] SearchResult[]
>(getKey, { >(getKey, {
revalidateFirstPage: true, revalidateFirstPage: false,
revalidateOnFocus: true, revalidateOnFocus: true,
revalidateAll: false, revalidateAll: false,
}); });