Page MenuHomeSoftware Heritage

D4890.diff
No OneTemporary

D4890.diff

diff --git a/swh/scheduler/journal_client.py b/swh/scheduler/journal_client.py
--- a/swh/scheduler/journal_client.py
+++ b/swh/scheduler/journal_client.py
@@ -48,10 +48,18 @@
}, f"Got unexpected {', '.join(set(messages) - set([msg_type]))} message types"
assert msg_type in messages, f"Expected {msg_type} messages"
- origin_visit_stats: Dict[Tuple[str, str], Dict] = {}
- for msg_dict in messages[msg_type]:
- if msg_dict["status"] in ("created", "ongoing"):
- continue
+ interesting_messages = [
+ msg for msg in messages[msg_type] if msg["status"] not in ("created", "ongoing")
+ ]
+
+ origin_visit_stats: Dict[Tuple[str, str], Dict] = {
+ (visit_stats.url, visit_stats.visit_type): attr.asdict(visit_stats)
+ for visit_stats in scheduler.origin_visit_stats_get(
+ list(set((vs["origin"], vs["type"]) for vs in interesting_messages))
+ )
+ }
+
+ for msg_dict in interesting_messages:
origin = msg_dict["origin"]
visit_type = msg_dict["type"]
empty_object = {
@@ -65,11 +73,7 @@
}
pk = origin, visit_type
if pk not in origin_visit_stats:
- visit_stats = scheduler.origin_visit_stats_get([pk])
- origin_visit_stats[pk] = (
- attr.asdict(visit_stats[0]) if visit_stats else empty_object
- )
-
+ origin_visit_stats[pk] = empty_object
visit_stats_d = origin_visit_stats[pk]
if msg_dict["status"] == "not_found":

File Metadata

Mime Type
text/plain
Expires
Dec 21 2024, 4:05 AM (11 w, 4 d ago)
Storage Engine
blob
Storage Format
Raw Data
Storage Handle
3221976

Event Timeline