diff --git a/project/server/main/feed.py b/project/server/main/feed.py
index 85a28e0..824aec7 100644
--- a/project/server/main/feed.py
+++ b/project/server/main/feed.py
@@ -10,7 +10,7 @@
 from project.server.main.aurehal import harvest_and_save_aurehal
 from project.server.main.logger import get_logger
 from project.server.main.parse import get_aurehal_from_OS, parse_hal
-from project.server.main.utils_swift import get_objects_by_prefix, upload_object
+from project.server.main.utils_swift import get_objects, get_paths_by_prefix, upload_object
 
 logger = get_logger(__name__)
 
@@ -179,18 +179,24 @@ def load_collection_from_object_storage(collection_name: str) -> None:
     logger.debug(f'dropping {collection_name} collection before insertion')
     myclient = pymongo.MongoClient('mongodb://mongo:27017/')
     myclient['hal'][collection_name].drop()
-    # 2. Collect all files from Object Storage
-    publications = get_objects_by_prefix(container='hal', prefix=f'{collection_name}/parsed/hal_parsed')
-    # 3. Extract oa_details from publications
-    oa_details_data = []
-    for publication in publications:
-        result = {
-            'hal_id': publication.get('hal_id'),
-            'oa_details': publication.get('oa_details')
-        }
-        oa_details_data.append(result)
-    # 4. Save it into mongo collection
-    current_file_oa_details = f'hal_oa_details.json'
-    json.dump(oa_details_data, open(current_file_oa_details, 'w'))
-    insert_data(collection_name=collection_name, output_file=current_file_oa_details)
-    return
\ No newline at end of file
+    # 2. Collect all paths from Object Storage container with prefix
+    paths = get_paths_by_prefix(container='hal', prefix=f'{collection_name}/parsed/hal_parsed')
+    logger.debug(f'{len(paths)} paths retrieved in the container with prefix')
+    for path in paths:
+        # 3. For each path, collect all objects
+        publications = get_objects(container='hal', path=path)
+        # publications = [item for sublist in objects for item in sublist]
+        oa_details_data = []
+        # 4. Extract oa_details from publications
+        for publication in publications:
+            result = {
+                'hal_id': publication.get('hal_id'),
+                'oa_details': publication.get('oa_details')
+            }
+            oa_details_data.append(result)
+        # 5. Save it into mongo collection
+        current_file_oa_details = f'hal_oa_details.json'
+        json.dump(oa_details_data, open(current_file_oa_details, 'w'))
+        insert_data(collection_name=collection_name, output_file=current_file_oa_details)
+        os.system(f'rm -rf {current_file_oa_details}')
+    return
diff --git a/project/server/main/utils_swift.py b/project/server/main/utils_swift.py
index 3048192..2260f82 100644
--- a/project/server/main/utils_swift.py
+++ b/project/server/main/utils_swift.py
@@ -47,7 +47,7 @@ def get_connection() -> swiftclient.Connection:
 
 
 @retry(delay=2, tries=50)
-def upload_object(container: str, source: str, target:str) -> str:
+def upload_object(container: str, source: str, target: str) -> str:
     logger.debug(f'Uploading {source} in {container} as {target}')
     cmd = init_cmd + f' upload {container} {source} --object-name {target}' \
                      f' --segment-size 1048576000 --segment-threads 100'
@@ -61,29 +61,29 @@ def download_object(container: str, filename: str, out: str) -> None:
     cmd = init_cmd + f' download {container} {filename} -o {out}'
     os.system(cmd)
 
+
 @retry(delay=2, tries=50)
 def get_objects(container: str, path: str) -> list:
     try:
         connection = get_connection()
-        df = pd.read_json(BytesIO(connection.get_object(container, path)[1]), compression='gzip')
+        df = pd.read_json(BytesIO(connection.get_object(
+            container, path)[1]), compression='gzip')
     except:
         df = pd.DataFrame([])
     return df.to_dict('records')
 
+
 @retry(delay=2, tries=50)
-def get_objects_by_prefix(container: str, prefix: str) -> list:
-    logger.debug(f'Retrieving object from container {container} and prefix {prefix}')
-    objects = []
+def get_paths_by_prefix(container: str, prefix: str) -> list:
+    logger.debug(
+        f'Retrieving paths from container {container} and prefix {prefix}')
     marker = None
     keep_going = True
+    filenames = []
     while keep_going:
         connection = get_connection()
-        content = connection.get_container(container=container, marker=marker, prefix=prefix)[1]
-        filenames = [file['name'] for file in content]
-        objects += [get_objects(container=container, path=filename) for filename in filenames]
+        content = connection.get_container(
+            container=container, marker=marker, prefix=prefix)[1]
+        filenames += [file['name'] for file in content]
         keep_going = len(content) == SWIFT_SIZE
-        if len(content) > 0:
-            marker = content[-1]['name']
-            logger.debug(f'Now {len(objects)} objects and counting')
-    flat_list = [item for sublist in objects for item in sublist]
-    return flat_list
\ No newline at end of file
+    return filenames
diff --git a/requirements.txt b/requirements.txt
index 0b9c6bb..1124e1e 100644
--- a/requirements.txt
+++ b/requirements.txt
@@ -3,19 +3,18 @@ Flask-Bootstrap==3.3.7.1
 Flask-Testing==0.7.1
 Flask-WTF==0.14.2
 gunicorn==20.0.4
-pymongo==3.8.0
+lxml==4.6.3
 pandas==1.2.5
+pycountry==20.7.3
+pymongo==3.8.0
 python-dateutil~=2.8.1
 python-keystoneclient==4.0.0
 python-swiftclient==3.9.0
 redis==3.5.3
-rq==1.9.0
 regex==2017.4.5
 requests==2.20.0
-lxml==4.6.3
-python-dateutil~=2.8.1
-pycountry==20.7.3
 retry~=0.9.2
+rq==1.9.0
+SPARQLWrapper==1.8.5
 tokenizers==0.10.1
 Unidecode==1.3.2
-SPARQLWrapper==1.8.5