Update app.py

app.py

@@ -225,7 +225,43 @@ def custom_css():
     """
 
 
-	
+#####################################################################
+# Antwort des Vektorstores in ein Dictionary packen
+def parse_vectorstore_response(response_text):
+    # Trenne die verschiedenen Einträge anhand des Musters
+    entries = re.split(r"(\d+\. page_content=)", response_text)
+    
+    # Entferne das erste leere Element, falls vorhanden
+    if entries[0] == '':
+        entries.pop(0)
+    
+    # Kombiniere die Identifier und Inhalte wieder zu vollständigen Einträgen
+    entries = [entries[i] + entries[i + 1] for i in range(0, len(entries), 2)]
+    
+    parsed_entries = []
+
+    for entry in entries:
+        # Extrahiere page_content
+        page_content_match = re.search(r"page_content='(.*?)' metadata=", entry, re.DOTALL)
+        page_content = page_content_match.group(1) if page_content_match else ''
+
+        # Extrahiere metadata
+        metadata_match = re.search(r"metadata=\{(.*?)\}", entry)
+        metadata_str = metadata_match.group(1) if metadata_match else ''
+        
+        # Konvertiere metadata in ein Dictionary
+        metadata = {}
+        for item in metadata_str.split(', '):
+            key, value = item.split(': ')
+            metadata[key.strip("'")] = value.strip("'")
+        
+        parsed_entries.append({
+            "page_content": page_content,
+            "metadata": metadata
+        })
+    
+    return parsed_entries
+    
 
 def create_history_entry(page_content, metadata):
     source = metadata.get('source', 'No source available')
@@ -275,8 +311,10 @@ def generate_auswahl(prompt_in, file, file_history, chatbot, history, anzahl_doc
     
             #File hochladen in Chroma und dann Antwort generieren
             results = generate_text_zu_doc(neu_file, prompt, k, rag_option, chatbot, history, vektordatenbank)
+
+        results_dictionary = parse_vectorstore_response(results)
             
-        for text in results:
+        for text in results_dictionary:
             print("text..............."+str(text))
             entry = create_history_entry(text['page_content'], text['metadata'])
             print("hier...........................")