Write only successful retrievals (i.e. ones that don't cause an exception) to WARC

qwarc/__init__.py

@@ -91,6 +91,8 @@ class Item:
 					exc = e # Pass the exception outward for the history
 				else:
 					action, writeToWarc = await responseHandler(url, attempt, response, None)
+				if response and exc is None and writeToWarc:
+					self.warc.write_client_response(response)
 				history.append((response, exc))
 				if action in (ACTION_SUCCESS, ACTION_IGNORE):
 					return response, tuple(history)
@@ -111,8 +113,6 @@ class Item:
 					pass
 			finally:
 				if response:
-					if writeToWarc:
-						self.warc.write_client_response(response)
 					await response.release()
 
 	async def process(self):

qwarc/utils.py

@@ -144,6 +144,7 @@ async def handle_response_default(url, attempt, response, exc):
 	Parameters: url (yarl.URL instance), attempt (int), response (aiohttp.ClientResponse or None), exc (Exception or None)
 		At least one of response and exc is not None.
 	Returns: (one of the qwarc.RESPONSE_* constants, bool signifying whether to write to WARC or not)
+		The latter is ignored when exc is not None; responses that triggered an exception are never written to WARC.
 	'''
 
 	#TODO: Document that `attempt` is reset on redirects