Update utils.py

utils.py

@@ -485,7 +485,7 @@ def document_storage_chroma(splits):
     vectorstore = Chroma.from_documents(documents=splits, embedding=embedding_fn) 
     
     return vectorstore
-"""
+
 ########################################################
 #Splits für den Vektorstore speichern - bzw. laden
 def save_splits(preprocessed_splits, original_splits, directory="chroma/kkg", preprocessed_filename="preprocessed_splits.pkl", original_filename="original_splits.pkl"):
@@ -506,7 +506,7 @@ def save_splits(preprocessed_splits, original_splits, directory="chroma/kkg", pr
     # Hochladen der Splits-Dateien zum Hugging Face Space
     upload_file_to_huggingface(preprocessed_filepath, f"{directory}/{preprocessed_filename}")
     upload_file_to_huggingface(original_filepath, f"{directory}/{original_filename}")
-
+"""
 def load_splits(directory="chroma/kkg", preprocessed_filename="preprocessed_splits.pkl", original_filename="original_splits.pkl"):
     # Vollständigen Pfad zur Datei erstellen
     preprocessed_filepath = os.path.join(directory, preprocessed_filename)
@@ -522,11 +522,41 @@ def load_splits(directory="chroma/kkg", preprocessed_filename="preprocessed_spli
         
         return preprocessed_splits, original_splits
     return None, None
+"""
+def load_splits(directory="chroma/kkg", preprocessed_filename="preprocessed_splits.pkl", original_filename="original_splits.pkl"):
+    preprocessed_splits = None
+    original_splits = None
+
+    try:
+        # Laden der vorverarbeiteten Splits aus dem Hugging Face Repository
+        preprocessed_file = hf_hub_download(
+            repo_id=STORAGE_REPO_ID,
+            filename=f"{directory}/{preprocessed_filename}",
+            repo_type="space",
+            token=hf_token
+        )
+        with open(preprocessed_file, "rb") as f:
+            preprocessed_splits = pickle.load(f)
+
+        # Laden der originalen Splits aus dem Hugging Face Repository
+        original_file = hf_hub_download(
+            repo_id=STORAGE_REPO_ID,
+            filename=f"{directory}/{original_filename}",
+            repo_type="space",
+            token=hf_token
+        )
+        with open(original_file, "rb") as f:
+            original_splits = pickle.load(f)
+
+    except Exception as e:
+        print(f"Fehler beim Laden der Splits: {str(e)}")
+
+    return preprocessed_splits, original_splits
 
 ########################################################
 #Vektorstore speichern - bzw. laden       
 #Laden des Vektorstores - aus den gespeicherten splits
-
+"""
 def load_vectorstore():
     splits_and_metadata = load_splits_and_metadata()
     if splits_and_metadata is not None:
@@ -549,7 +579,7 @@ def save_split_to_original_mapping(mapping, directory="chroma/kkg", filename="ma
     
     # Hochladen der Mapping-Datei zum Hugging Face Space
     upload_file_to_huggingface(filepath, f"{directory}/{filename}")
-
+"""
 def load_split_to_original_mapping(directory="chroma/kkg", filename="mapping.pkl"):
     # Vollständigen Pfad zur Datei erstellen
     filepath = os.path.join(directory, filename)
@@ -559,7 +589,24 @@ def load_split_to_original_mapping(directory="chroma/kkg", filename="mapping.pkl
         with open(filepath, "rb") as f:
             return pickle.load(f)
     return None
+"""
 
+def load_split_to_original_mapping(directory="chroma/kkg", filename="mapping.pkl"):
+    try:
+        # Laden des Mappings aus dem Hugging Face Repository
+        file_path = hf_hub_download(
+            repo_id=STORAGE_REPO_ID,
+            filename=f"{directory}/{filename}",
+            repo_type="space",
+            token=hf_token
+        )
+        
+        with open(file_path, "rb") as f:
+            return pickle.load(f)
+    
+    except Exception as e:
+        print(f"Fehler beim Laden des Mappings: {str(e)}")
+        return None
 
 
 #######################################