Update app.py

app.py

@@ -41,8 +41,11 @@ def generate_random_string(length):
 #if "pdf_files" not in st.session_state:
     #st.session_state.pdf_files = None
 
-if "documents" not in st.session_state:
-    st.session_state.documents = None
+#if "documents" not in st.session_state:
+#    st.session_state.documents = None
+
+if "query_engine" not in st.session_state:
+    st.session_state.query_engine = None
 
 with st.sidebar:
     st.subheader("Upload your Documents Here: ")
@@ -64,57 +67,45 @@ with st.sidebar:
         try:
             start_1 = timeit.default_timer() # Start timer
             st.write(f"QA文档加载开始：{start_1}")
-            st.session_state.documents = SimpleDirectoryReader(uploadedfile_path).load_data()
+            #st.session_state.documents = SimpleDirectoryReader(uploadedfile_path).load_data()
+            documents = SimpleDirectoryReader(uploadedfile_path).load_data()
             end_1 = timeit.default_timer() # Start timer
             st.write(f"QA文档加载结束：{end_1}")
             st.write(f"QA文档加载耗时：{end_1 - start_1}")
         except Exception as e:
             print("文档加载出现问题/Waiting for path creation.")
-    
-start_2 = timeit.default_timer() # Start timer
-st.write(f"向量模型加载开始：{start_2}")
-if "embed_model" not in st.session_state:
-    st.session_state.embed_model = LangchainEmbedding(HuggingFaceEmbeddings(model_name='sentence-transformers/all-MiniLM-L6-v2'))
-end_2 = timeit.default_timer() # Start timer
-st.write(f"向量模型加载加载结束：{end_2}")
-st.write(f"向量模型加载耗时：{end_2 - start_2}")
-
-if "llm_predictor" not in st.session_state:
-    st.session_state.llm_predictor = LLMPredictor(HuggingFaceHub(repo_id="HuggingFaceH4/starchat-beta", model_kwargs={"min_length":100, "max_new_tokens":1024, "do_sample":True, "temperature":0.2,"top_k":50, "top_p":0.95, "eos_token_id":49155}))
-
-if "service_context" not in st.session_state:
-    st.session_state.service_context = ServiceContext.from_defaults(llm_predictor=st.session_state.llm_predictor, embed_model=st.session_state.embed_model)
-
-start_3 = timeit.default_timer() # Start timer
-st.write(f"向量库构建开始：{start_3}")
-if "new_index" not in st.session_state:
-    st.session_state.new_index = VectorStoreIndex.from_documents(
-    st.session_state.documents,
-    service_context=st.session_state.service_context,
-)
-end_3 = timeit.default_timer() # Start timer
-st.write(f"向量库构建结束：{end_3}")
-st.write(f"向量库构建耗时：{end_3 - start_3}")
-
-if "directory_path" not in st.session_state:
-    st.session_state.directory_path = generate_random_string(20)
-    os.makedirs(st.session_state.directory_path)
-    
-st.session_state.new_index.storage_context.persist("st.session_state.directory_path")
-
-if "storage_context" not in st.session_state:
-    st.session_state.storage_context = StorageContext.from_defaults(persist_dir="st.session_state.directory_path")
-
-start_4 = timeit.default_timer() # Start timer
-st.write(f"向量库装载开始：{start_4}")
-if "loadedindex" not in st.session_state:
-    st.session_state.loadedindex = load_index_from_storage(storage_context=st.session_state.storage_context, service_context=st.session_state.service_context)
-end_4 = timeit.default_timer() # Start timer
-st.write(f"向量库装载结束：{end_4}")
-st.write(f"向量库装载耗时：{end_4 - start_4}")
-
-if "query_engine" not in st.session_state:
-    st.session_state.query_engine = st.session_state.loadedindex.as_query_engine()
+        start_2 = timeit.default_timer() # Start timer
+        st.write(f"向量模型加载开始：{start_2}")
+        if "embed_model" not in st.session_state:
+            st.session_state.embed_model = LangchainEmbedding(HuggingFaceEmbeddings(model_name='sentence-transformers/all-MiniLM-L6-v2'))
+        end_2 = timeit.default_timer() # Start timer
+        st.write(f"向量模型加载结束：{end_2}")
+        st.write(f"向量模型加载耗时：{end_2 - start_2}")
+        if "llm_predictor" not in st.session_state:
+            st.session_state.llm_predictor = LLMPredictor(HuggingFaceHub(repo_id="HuggingFaceH4/starchat-beta", model_kwargs={"min_length":100, "max_new_tokens":1024, "do_sample":True, "temperature":0.2,"top_k":50, "top_p":0.95, "eos_token_id":49155}))
+        if "service_context" not in st.session_state:
+            st.session_state.service_context = ServiceContext.from_defaults(llm_predictor=st.session_state.llm_predictor, embed_model=st.session_state.embed_model)
+        start_3 = timeit.default_timer() # Start timer
+        st.write(f"向量库构建开始：{start_3}")
+        if "new_index" not in st.session_state:
+            st.session_state.new_index = VectorStoreIndex.from_documents(st.session_state.documents, service_context=st.session_state.service_context)
+        end_3 = timeit.default_timer() # Start timer
+        st.write(f"向量库构建结束：{end_3}")
+        st.write(f"向量库构建耗时：{end_3 - start_3}")
+        if "directory_path" not in st.session_state:
+            st.session_state.directory_path = generate_random_string(20)
+            os.makedirs(st.session_state.directory_path)
+        st.session_state.new_index.storage_context.persist("st.session_state.directory_path")
+        if "storage_context" not in st.session_state:
+            st.session_state.storage_context = StorageContext.from_defaults(persist_dir="st.session_state.directory_path")
+        start_4 = timeit.default_timer() # Start timer
+        st.write(f"向量库装载开始：{start_4}")
+        if "loadedindex" not in st.session_state:
+            st.session_state.loadedindex = load_index_from_storage(storage_context=st.session_state.storage_context, service_context=st.session_state.service_context)
+        end_4 = timeit.default_timer() # Start timer
+        st.write(f"向量库装载结束：{end_4}")
+        st.write(f"向量库装载耗时：{end_4 - start_4}")
+        st.session_state.query_engine = st.session_state.loadedindex.as_query_engine()
     
 user_question = st.text_input("Enter your query:")
 if user_question !="" and not user_question.strip().isspace() and not user_question == "" and not user_question.strip() == "" and not user_question.isspace():