Fix running 'cleanup' method on extractors that fail to start

src/auto_archiver/core/module.py

@@ -277,6 +277,9 @@ class LazyBaseModule:
         # finally, get the class instance
         instance: BaseModule = getattr(sys.modules[sub_qualname], class_name)()
 
+        # save the instance for future easy loading
+        self._instance = instance
+
         # set the name, display name and module factory
         instance.name = self.name
         instance.display_name = self.display_name
@@ -289,8 +292,6 @@ class LazyBaseModule:
         instance.config_setup(config)
         instance.setup()
 
-        # save the instance for future easy loading
-        self._instance = instance
         return instance
 
     def __repr__(self):

src/auto_archiver/core/orchestrator.py

@@ -387,8 +387,10 @@ Here's how that would look: \n\nsteps:\n  extractors:\n  - [your_extractor_name_
                 except (KeyboardInterrupt, Exception) as e:
                     if not isinstance(e, KeyboardInterrupt) and not isinstance(e, SetupError):
                         logger.error(f"Error during setup of modules: {e}\n{traceback.format_exc()}")
-                    if loaded_module and module_type == "extractor":
-                        loaded_module.cleanup()
+
+                    # access the _instance here because loaded_module may not return if there's an error
+                    if lazy_module._instance and module_type == "extractor":
+                        lazy_module._instance.cleanup()
                     raise e
 
                 if not loaded_module:

tests/data/test_modules/example_extractor/example_extractor.py

@@ -1,6 +1,11 @@
 from auto_archiver.core import Extractor
 
+from loguru import logger
+
 
 class ExampleExtractor(Extractor):
     def download(self, item):
-        print("download")
+        logger.info("download")
+
+    def cleanup(self):
+        logger.info("cleanup")

tests/data/test_modules/example_module/example_module.py

@@ -1,27 +1,29 @@
 from auto_archiver.core import Extractor, Enricher, Feeder, Database, Storage, Formatter, Metadata
 
+from loguru import logger
+
 
 class ExampleModule(Extractor, Enricher, Feeder, Database, Storage, Formatter):
     def download(self, item):
-        print("download")
+        logger.info("download")
 
     def __iter__(self):
         yield Metadata().set_url("https://example.com")
 
     def done(self, result):
-        print("done")
+        logger.info("done")
 
     def enrich(self, to_enrich):
-        print("enrich")
+        logger.info("enrich")
 
     def get_cdn_url(self, media):
         return "nice_url"
 
     def save(self, item):
-        print("save")
+        logger.info("save")
 
     def uploadf(self, file, key, **kwargs):
-        print("uploadf")
+        logger.info("uploadf")
 
     def format(self, item):
-        print("format")
+        logger.info("format")

tests/test_orchestrator.py

@@ -237,3 +237,23 @@ def test_wrong_step_type(test_args, caplog):
     with pytest.raises(SetupError) as err:
         orchestrator.setup(args)
         assert "Module 'example_extractor' is not a feeder" in str(err.value)
+
+
+def test_load_failed_extractor_cleanup(test_args, mocker, caplog):
+    orchestrator = ArchivingOrchestrator()
+
+    # hack to set up the paths so we can patch properly
+    orchestrator.module_factory.setup_paths([TEST_MODULES])
+
+    # patch example_module.setup to throw an exception
+    mocker.patch(
+        "auto_archiver.modules.example_extractor.example_extractor.ExampleExtractor.setup",
+        side_effect=Exception("Test exception"),
+    )
+
+    with pytest.raises(Exception):
+        orchestrator.setup(test_args + ["--extractors", "example_extractor"])
+
+    assert "Error during setup of modules: Test exception" in caplog.text
+    # make sure the 'cleanup' is called
+    assert "cleanup" in caplog.text