Update app.py

app.py

@@ -1,37 +1,32 @@
 import os
 import gradio as gr
-from langchain.document_loaders import PyPDFLoader
+from langchain_community.document_loaders import PyPDFLoader
 from langchain.text_splitter import RecursiveCharacterTextSplitter
-from langchain.embeddings import HuggingFaceEmbeddings
-from langchain.vectorstores import FAISS
+from langchain_community.embeddings import HuggingFaceEmbeddings
+from langchain_community.vectorstores import FAISS
 from langchain.chains import ConversationalRetrievalChain
 from langchain.memory import ConversationBufferMemory
-from langchain.llms import HuggingFacePipeline
+from langchain_community.llms import HuggingFacePipeline
 from transformers import pipeline
 
-# **Embeddings-Modell (kein API-Key nötig, lokal geladen)**
 EMBEDDINGS_MODEL_NAME = "sentence-transformers/all-MiniLM-L6-v2"
-LLM_MODEL_NAME = "google/flan-t5-small"  # Alternativ: "google/flan-t5-base", etc.
+LLM_MODEL_NAME = "google/flan-t5-small"
 
-# **Dokumente laden und aufteilen**
 def load_and_split_docs(list_file_path):
+    if not list_file_path:
+        return [], "Fehler: Keine Dokumente gefunden!"
     loaders = [PyPDFLoader(x) for x in list_file_path]
     documents = []
     for loader in loaders:
         documents.extend(loader.load())
     text_splitter = RecursiveCharacterTextSplitter(chunk_size=512, chunk_overlap=32)
-    doc_splits = text_splitter.split_documents(documents)
-    return doc_splits
+    return text_splitter.split_documents(documents)
 
-# **Vektor-Datenbank mit FAISS erstellen**
 def create_db(docs):
     embeddings = HuggingFaceEmbeddings(model_name=EMBEDDINGS_MODEL_NAME)
-    faiss_index = FAISS.from_documents(docs, embeddings)
-    return faiss_index
+    return FAISS.from_documents(docs, embeddings)
 
-# **LLM-Kette initialisieren**
 def initialize_llm_chain(llm_model, temperature, max_tokens, vector_db):
-    # Hugging Face Pipeline lokal verwenden
     local_pipeline = pipeline(
         "text2text-generation",
         model=llm_model,
@@ -39,37 +34,15 @@ def initialize_llm_chain(llm_model, temperature, max_tokens, vector_db):
         temperature=temperature
     )
     llm = HuggingFacePipeline(pipeline=local_pipeline)
-    memory = ConversationBufferMemory(memory_key="chat_history", return_messages=True)
+    memory = ConversationBufferMemory(memory_key="chat_history")
     retriever = vector_db.as_retriever()
-    
-    # Retrieval-Augmented QA-Kette
-    qa_chain = ConversationalRetrievalChain.from_llm(
+    return ConversationalRetrievalChain.from_llm(
         llm,
         retriever=retriever,
         memory=memory,
         return_source_documents=True
     )
-    return qa_chain
 
-# **Datenbank und Kette initialisieren**
-def initialize_database(list_file_obj):
-    list_file_path = [x.name for x in list_file_obj if x is not None]
-    doc_splits = load_and_split_docs(list_file_path)
-    vector_db = create_db(doc_splits)
-    return vector_db, "Datenbank erfolgreich erstellt!"
-
-def initialize_llm_chain_wrapper(llm_temperature, max_tokens, vector_db):
-    qa_chain = initialize_llm_chain(LLM_MODEL_NAME, llm_temperature, max_tokens, vector_db)
-    return qa_chain, "QA-Chatbot ist bereit!"
-
-# **Konversation mit QA-Kette führen**
-def conversation(qa_chain, message, history):
-    response = qa_chain({"question": message, "chat_history": history})
-    response_text = response["answer"]
-    sources = [doc.metadata["source"] for doc in response["source_documents"]]
-    return qa_chain, response_text, history + [(message, response_text)]
-
-# **Gradio-Benutzeroberfläche**
 def demo():
     with gr.Blocks() as demo:
         vector_db = gr.State()
@@ -86,15 +59,15 @@ def demo():
                 qachain_btn = gr.Button("Initialisiere QA-Chatbot")
 
             with gr.Column():
-                chatbot = gr.Chatbot(height=400)
+                chatbot = gr.Chatbot(type='messages', height=400)
                 msg = gr.Textbox(placeholder="Frage eingeben...")
                 submit_btn = gr.Button("Absenden")
 
         db_btn.click(initialize_database, [document], [vector_db, db_status])
         qachain_btn.click(initialize_llm_chain_wrapper, [slider_temperature, slider_max_tokens, vector_db], [qa_chain])
-        submit_btn.click(conversation, [qa_chain, msg, chatbot], [qa_chain, msg, chatbot])
+        submit_btn.click(conversation, [qa_chain, msg, []], [qa_chain, "message", "history"])
 
-    demo.launch(debug=True)
+    demo.launch(debug=True, enable_queue=True)
 
 if __name__ == "__main__":
     demo()