Fix to dataset extractor

backend/app/rabbitmq/listeners.py

@@ -56,7 +56,6 @@ async def submit_file_job(
     parameters: dict,
     user: UserOut,
     rabbitmq_client: BlockingChannel,
-    token: str = Depends(get_token),
 ):
     # Create an entry in job history with unique ID
     job = EventListenerJobDB(

backend/app/routers/datasets.py

@@ -466,7 +466,6 @@ async def save_file(
     file: UploadFile = File(...),
     es=Depends(dependencies.get_elasticsearchclient),
     rabbitmq_client: BlockingChannel = Depends(dependencies.get_rabbitmq),
-    credentials: HTTPAuthorizationCredentials = Security(security),
     allow: bool = Depends(Authorization("uploader")),
 ):
     if (dataset := await DatasetDB.get(PydanticObjectId(dataset_id))) is not None:
@@ -485,14 +484,13 @@ async def save_file(
                     status_code=404, detail=f"Folder {folder_id} not found"
                 )
 
-        access_token = credentials.credentials
+        # access_token = credentials.credentials
         await add_file_entry(
             new_file,
             user,
             fs,
             es,
             rabbitmq_client,
-            access_token,
             file.file,
             content_type=file.content_type,
         )
@@ -583,8 +581,8 @@ async def download_dataset(
     if (dataset := await DatasetDB.get(PydanticObjectId(dataset_id))) is not None:
         current_temp_dir = tempfile.mkdtemp(prefix="rocratedownload")
         crate = ROCrate()
-        user_full_name = user.first_name + " " + user.last_name
-        user_crate_id = str(user.id)
+        user_full_name = user["first_name"] + " " + user["last_name"]
+        user_crate_id = str(user["id"])
         crate.add(Person(crate, user_crate_id, properties={"name": user_full_name}))
 
         manifest_path = os.path.join(current_temp_dir, "manifest-md5.txt")
@@ -676,7 +674,7 @@ async def download_dataset(
             f.write("Internal-Sender-Identifier: " + dataset_id + "\n")
             f.write("Internal-Sender-Description: " + dataset.description + "\n")
             f.write("Contact-Name: " + user_full_name + "\n")
-            f.write("Contact-Email: " + user.email + "\n")
+            f.write("Contact-Email: " + user["email"] + "\n")
         crate.add_file(
             bagit_path, dest_path="bagit.txt", properties={"name": "bagit.txt"}
         )

backend/app/routers/feeds.py

@@ -43,7 +43,6 @@ async def check_feed_listeners(
     file_out: FileOut,
     user: UserOut,
     rabbitmq_client: BlockingChannel,
-    token: str,
 ):
     """Automatically submit new file to listeners on feeds that fit the search criteria."""
     listener_ids_found = []
@@ -64,7 +63,6 @@ async def check_feed_listeners(
                 {},  # parameters
                 user,
                 rabbitmq_client,
-                token,
             )
     return listener_ids_found
 

backend/app/routers/files.py

@@ -92,7 +92,6 @@ async def add_file_entry(
     fs: Minio,
     es: Elasticsearch,
     rabbitmq_client: BlockingChannel,
-    token: str,
     file: Optional[io.BytesIO] = None,
     content_type: Optional[str] = None,
 ):
@@ -111,7 +110,7 @@ async def add_file_entry(
 
     await new_file.insert()
     new_file_id = new_file.id
-    content_type_obj = get_content_type(content_type, file)
+    content_type_obj = get_content_type(new_file.name, content_type)
 
     # Use unique ID as key for Minio and get initial version ID
     response = fs.put_object(
@@ -146,7 +145,10 @@ async def add_file_entry(
 
     # Submit file job to any qualifying feeds
     await check_feed_listeners(
-        es, FileOut(**new_file.dict()), user, rabbitmq_client, token
+        es,
+        FileOut(**new_file.dict()),
+        user,
+        rabbitmq_client,
     )
 
 

backend/app/routers/utils.py

@@ -6,7 +6,8 @@
 
 
 def get_content_type(
-    content_type: Optional[str] = None, file: Optional[io.BytesIO] = None
+    filename: str,
+    content_type: Optional[str] = None,
 ):
     """Returns ContentType object given a content_tyoe, also guessed the content_type if none is provided
 
@@ -16,7 +17,7 @@ def get_content_type(
     """
 
     if content_type is None:
-        content_type = mimetypes.guess_type(file.name)
+        content_type = mimetypes.guess_type(filename)
         content_type = content_type[0] if len(content_type) > 1 else content_type
     type_main = content_type.split("/")[0] if type(content_type) is str else "N/A"
     return ContentType(content_type=content_type, main_type=type_main)

backend/app/routers/visualization.py

@@ -60,7 +60,7 @@ async def add_Visualization(
     )
 
     await visualization_db.insert()
-    visualization_db.content_type = get_content_type(file.content_type, file.file)
+    visualization_db.content_type = get_content_type(file.filename, file.content_type)
     visualization_id = visualization_db.id
 
     # Use unique ID as key for Minio