Update utils.py

utils.py

@@ -101,6 +101,25 @@ HF_WRITE = os.getenv("HF_WRITE")
 # HfApi-Instanz erstellen
 api = HfApi()
 
+
+
+# Login mit dem Token
+login(token=hf_token)
+
+# API-Instanz erstellen
+api = HfApi()
+
+# Überprüfen, ob das Repository existiert und zugänglich ist
+try:
+    repo_info = api.list_repo_files(repo_id=STORAGE_REPO_ID, repo_type=REPO_TYPE)
+    print(f"Repository '{STORAGE_REPO_ID}' enthält folgende Dateien: {repo_info}")
+except Exception as e:
+    print(f"Fehler beim Zugriff auf das Repository: {e}")
+
+
+
+
+
 #Maoing für die Splits (orginal und Preprocessed
 split_to_original_mapping = []
 
@@ -331,23 +350,6 @@ def split_documents_with_id(docs, text_splitter):
             splits.append(split_doc)
     return splits
 
-#######################################
-# Dokumente aus anderem Space laden
-#######################################
-#ein File aus dem Space mit der REPO_ID laden
-def download_file_from_hf(file_name, save_path):
-    url = f"https://huggingface.co/{STORAGE_REPO_ID}/resolve/main/{file_name}"
-    response = requests.get(url)
-    response.raise_for_status()  # Raise an error for bad status codes
-    with open(save_path, 'wb') as file:
-        file.write(response.content)
-    return save_path
-
-#Liste aller Files in dem Space mit der Repo_id
-def list_files_in_hf_repo(repo_id):
-    repo_info = api.list_repo_files(repo_id=repo_id)
-    return repo_info
-
     
 
 ########################################
@@ -523,6 +525,23 @@ def upload_file_to_huggingface(file_path, upload_path):
         repo_id=STORAGE_REPO_ID,
         repo_type=REPO_TYPE,
         token=HF_WRITE
+
+#######################################
+# Dokumente aus anderem Space laden
+#######################################
+#ein File aus dem Space mit der REPO_ID laden
+def download_file_from_hf(file_name, save_path):
+    url = f"https://huggingface.co/{STORAGE_REPO_ID}/resolve/main/{file_name}"
+    headers = {"Authorization": f"Bearer {HF_READ}"}
+    response = requests.get(url, headers=headers)
+    response.raise_for_status()  # Raise an error for bad status codes
+    with open(save_path, 'wb') as file:
+        file.write(response.content)
+    return save_path
+
+def list_files_in_hf_repo(repo_id):
+    repo_info = api.list_repo_files(repo_id=repo_id, repo_type=REPO_TYPE)
+    return repo_info
     )