Preflight plugin dependencies before crawl hooks

archivebox/services/runner.py

@@ -817,7 +817,17 @@ class CrawlRunner:
             + CrawlCompletedEvent.model_fields["event_timeout"].default
             + 30.0
         )
-        await plan.seed_config(self.bus)
+        if plan.catalog:
+            await abx_install_plugins(
+                plan,
+                output_dir=output_dir,
+                emit_jsonl=False,
+                bus=self.bus,
+                BinaryService=None,
+                ProcessService=None,
+            )
+        else:
+            await plan.seed_config(self.bus)
         plan.attach_services(
             self.bus,
             url=snapshot["url"],
@@ -834,6 +844,7 @@ class CrawlRunner:
             emit_jsonl=False,
             abort_requested=self.crawl_is_cancelled,
             PluginBinariesService=None,
+            PluginBinaryEnvService=None,
             BinaryService=None,
             ProcessService=None,
             ArchiveResultService=None,

archivebox/tests/test_crawl_runner.py

@@ -657,6 +657,35 @@ def test_crawl_runner_empty_plugin_selection_emits_lifecycle_and_seals_crawl(tmp
     assert snapshot.archiveresult_set.count() == 0
 
 
+@pytest.mark.django_db(transaction=True)
+def test_crawl_runner_preflights_plugins_before_crawl_lifecycle(tmp_path):
+    from abx_dl.events import CrawlEvent, CrawlSetupEvent, InstallEvent
+    from archivebox.base_models.models import get_or_create_system_user_pk
+    from archivebox.crawls.models import Crawl
+    from archivebox.services.runner import CrawlRunner
+
+    crawl = Crawl.objects.create(
+        urls="https://example.com",
+        config={
+            "ABXPKG_LIB_DIR": str(tmp_path / "lib"),
+            "PLUGINS": "base",
+            "CHROME_BINARY": "",
+        },
+        created_by_id=get_or_create_system_user_pk(),
+    )
+    runner = CrawlRunner(crawl)
+    event_order = []
+    runner.bus.on(InstallEvent, lambda _event: event_order.append(InstallEvent))
+    runner.bus.on(CrawlEvent, lambda _event: event_order.append(CrawlEvent))
+    runner.bus.on(CrawlSetupEvent, lambda _event: event_order.append(CrawlSetupEvent))
+
+    asyncio.run(runner.run())
+
+    assert event_order.count(InstallEvent) == 1
+    assert event_order.index(InstallEvent) < event_order.index(CrawlEvent)
+    assert event_order.index(InstallEvent) < event_order.index(CrawlSetupEvent)
+
+
 @pytest.mark.django_db(transaction=True)
 def test_crawl_runner_resolves_persona_and_crawl_config_for_each_live_snapshot():
     from abx_dl.events import SnapshotCompletedEvent

archivebox/tests/test_migrations_07_to_09.py

@@ -403,7 +403,7 @@ def test_update_saves_migrated_snapshots_without_foreign_key_errors(archive_07):
     result = run_archivebox_migration_cmd(work_dir, ["init"], timeout=45)
     assert result.returncode == 0, f"Init failed: {result.stderr}"
 
-    result = run_archivebox_migration_cmd(work_dir, ["update"], timeout=60)
+    result = run_archivebox_migration_cmd(work_dir, ["update", "--migrate-only"], timeout=60)
     output = result.stdout + result.stderr
     assert result.returncode == 0, f"Update failed after migration: {result.stderr}"
     assert "FOREIGN KEY constraint failed" not in output