Update run.py

run.py

@@ -2,7 +2,7 @@
 # Title:  Gradio Interface to LLM-chatbot with dynamic RAG-funcionality and ChromaDB
 # Author: Andreas Fischer
 # Date:   October 10th, 2024
-# Last update: October 11th, 2024
+# Last update: October 12th, 2024
 ##########################################################################################
 
 import os
@@ -108,23 +108,26 @@ def split_with_overlap(text,chunk_size=3500, overlap=700):
    chunks.append(text[i:end])
  return chunks
 
-def add_doc(path):
+
+def add_doc(path, session):
   print("def add_doc!")
   print(path)
   anhang=False
   if(str.lower(path).endswith(".pdf") and os.path.exists(path)):
       doc=convertPDF(path)
       if(len(doc[0])>5): 
-        gr.Info("PDF uploaded, start Indexing excerpt (first 5 pages)!")        
+        gr.Info("PDF uploaded to DB_"+str(session)+", start Indexing excerpt (first 5 pages)!")        
       else:
-        gr.Info("PDF uploaded, start Indexing!")            
+        gr.Info("PDF uploaded to DB_"+str(session)+", start Indexing!")            
       doc="\n\n".join(doc[0][0:5]) 
       anhang=True
-      
+  else:
+    gr.Info("No PDF attached - answer based on DB_"+str(session)+".")          
   client = chromadb.PersistentClient(path="output/general_knowledge")
   print(str(client.list_collections()))
   #global collection
-  dbName="test"
+  print(str(session))
+  dbName="DB_"+str(session)
   if(not "name="+dbName in str(client.list_collections())): 
     #  client.delete_collection(name=dbName) 
     collection = client.create_collection(
@@ -157,17 +160,28 @@ def add_doc(path):
   return(collection)
 
 #split_with_overlap("test me if you can",2,1)
+from datetime import date
+databases=[(date.today(),"0")] # list of all databases
 
 import gradio as gr
 import re  
-def multimodalResponse(message,history,dropdown):
+def multimodalResponse(message,history,dropdown, request: gr.Request):
   print("def multimodal response!")
+  global databases
+  if request:
+    session=request.session_hash
+  else:
+    session="0"
   length=str(len(history))
+  print(databases)
+  if(not databases[-1][1]==session):
+    databases.append((date.today(),session))
+    #print(databases)
   query=message["text"]
   if(len(message["files"])>0): # is there at least one file attached?
-    collection=add_doc(message["files"][0])
-  else:
-    collection=add_doc(message["text"])
+    collection=add_doc(message["files"][0], session)
+  else: # otherwise, you still want to get the collection with the session-based db
+    collection=add_doc(message["text"], session)
   client = chromadb.PersistentClient(path="output/general_knowledge")
   print(str(client.list_collections()))
   x=collection.get(include=[])["ids"]  
@@ -214,11 +228,3 @@ i.launch() #allowed_paths=["."])
 
 
 
-
-
-
-
-
-
-
-