Find snapshot completion events through bus history

archivebox/services/runner.py

@@ -583,14 +583,13 @@ class CrawlRunner:
                 )
                 emitted_snapshot_event = self.bus.emit(snapshot_event)
                 await emitted_snapshot_event.now()
-                await emitted_snapshot_event.event_results_list()
-                snapshot_event_results = getattr(snapshot_event, "event_results", {})
-                snapshot_event_children = getattr(snapshot_event, "event_children", [])
-                completed_snapshot = next(
-                    (child for child in snapshot_event_children if isinstance(child, SnapshotCompletedEvent)),
-                    None,
+                completed_snapshot = await self.bus.find(
+                    SnapshotCompletedEvent,
+                    child_of=emitted_snapshot_event,
+                    past=True,
+                    future=snapshot_phase_timeout,
                 )
-                if snapshot_event_results and completed_snapshot is None:
+                if completed_snapshot is None:
                     raise RuntimeError(f"Snapshot {snapshot_id} did not complete")
                 await self.enqueue_discovered_snapshots_from_outputs(snapshot)
             finally:

archivebox/tests/test_runner.py

@@ -28,9 +28,24 @@ class _DummyBus:
 
     def emit(self, event):
         self.emitted.append(event)
+        bus = self
 
         class _Pending:
+            def __getattr__(self, name):
+                return getattr(event, name)
+
             async def now(self, *args, **kwargs):
+                from abx_dl.events import SnapshotCompletedEvent, SnapshotEvent
+
+                if isinstance(event, SnapshotEvent):
+                    bus.emitted.append(
+                        SnapshotCompletedEvent(
+                            url=event.url,
+                            snapshot_id=event.snapshot_id,
+                            output_dir=event.output_dir,
+                            event_parent_id=event.event_id,
+                        ),
+                    )
                 return event
 
             async def wait(self, *args, **kwargs):
@@ -41,6 +56,17 @@ class _DummyBus:
 
         return _Pending()
 
+    async def find(self, event_type, where=None, child_of=None, **kwargs):
+        for event in reversed(self.emitted):
+            if not isinstance(event, event_type):
+                continue
+            if child_of is not None and event.event_parent_id != child_of.event_id:
+                continue
+            if where is not None and not where(event):
+                continue
+            return event
+        return None
+
     async def stop(self):
         return None
 

pyproject.toml

@@ -1,6 +1,6 @@
 [project]
 name = "archivebox"
-version = "0.9.30rc19"
+version = "0.9.30rc20"
 requires-python = ">=3.13"
 description = "Self-hosted internet archiving solution."
 authors = [{name = "Nick Sweeting", email = "pyproject.toml@archivebox.io"}]