better chat interface, added answer generation bot and fixed previous…

…ly mysteriously deleted files

backend/config.json

@@ -17,7 +17,7 @@
     "search_type" : "similarity",
     "reranking" : false,
     "long_context_reorder" : false,
-    "structured_query": true,
+    "structured_query": false,
     "use_chroma_for_saving_metadata": false,
     "chunk_size": 1000,
     "chroma_metadata_dir": "../data/chroma_db_metadata"

frontend/paths.json

@@ -10,5 +10,9 @@
     "structured_query": {
         "docker" : "http://fastapi:8081/structuredquery/",
         "local" : "http://0.0.0.0:8081/structuredquery/"
+    },
+    "documentation_query": {
+        "docker" : "http://fastapi:8083/documentationquery/",
+        "local" : "http://0.0.0.0:8083/documentationquery/"
     }
 }

frontend/ui.py

@@ -29,13 +29,17 @@
     config_path = Path("../backend/config.json")
     ui_loader = UILoader(config_path)
 
-query_type = st.selectbox("Select Query Type", ["Dataset", "Flow"])
-llm_filter = st.toggle("LLM Filter")
 # Chat input box
 user_input = ui_loader.chat_entry()
 
+ui_loader.create_chat_interface(None)
+query_type = st.selectbox("Select Query Type", ["General Query","Dataset", "Flow"], key="query_type_2")
+llm_filter = st.toggle("LLM Filter")
 # Chat interface
 if user_input:
     ui_loader.create_chat_interface(
         user_input, query_type=query_type, llm_filter=llm_filter
     )
+    ui_loader.query_type = st.selectbox("Select Query Type", ["General Query","Dataset", "Flow"], key="query_type_3")
+    ui_loader.llm_filter = st.toggle("LLM Filter", key="llm_filter_2")
+

frontend/ui_utils.py

@@ -41,9 +41,11 @@ def display_results(initial_response):
     """
     Description: Display the results in a DataFrame
     """
-    st.write("OpenML Agent: ")
-    st.dataframe(initial_response)
-
+    # st.write("OpenML Agent: ")
+    try:
+        st.dataframe(initial_response)
+    except:
+        st.write(initial_response)
 
 class LLMResponseParser:
     """
@@ -145,6 +147,22 @@ def fetch_llm_response(self, query: str):
                 f"{llm_response_path['local']}{query}"
             ).json()
         return self.llm_response
+
+    def fetch_documentation_query(self, query: str):
+        """
+        Description: Fetch the response for a general or documentation or code query from the LLM service as a JSON
+        """
+        documentation_response_path = self.paths["documentation_query"]
+        try:
+            self.documentation_response = requests.get(
+                f"{documentation_response_path['docker']}{query}",
+                json={"query": query},
+            ).json()
+        except:
+            self.documentation_response = requests.get(
+                f"{documentation_response_path['local']}{query}",
+                json={"query": query},
+            ).json()
 
     def fetch_structured_query(self, query_type: str, query: str):
         """
@@ -369,13 +387,17 @@ def chat_entry(self):
             self.chatbot_display, max_chars=self.chatbot_input_max_chars
         )
 
-    def create_chat_interface(self, user_input, query_type, llm_filter):
+    def create_chat_interface(self, user_input, query_type=None, llm_filter=None):
         """
         Description: Create the chat interface and display the chat history and results. Show the user input and the response from the OpenML Agent.
 
         """
         self.query_type = query_type
         self.llm_filter = llm_filter
+        if user_input is None:
+            with st.chat_message(name = "ai"):
+                st.write("OpenML Agent: ", "Hello! How can I help you today?")
+
         # Handle user input
         if user_input:
             st.session_state.messages.append({"role": "user", "content": user_input})
@@ -389,9 +411,11 @@ def create_chat_interface(self, user_input, query_type, llm_filter):
         # Display chat history
         for message in st.session_state.messages:
             if message["role"] == "user":
-                st.write(f"You: {message['content']}")
+                with st.chat_message(name = "user"):
+                    display_results(message["content"])
             else:
-                display_results(message["content"])
+                with st.chat_message(name = "ai"):
+                    display_results(message["content"])
 
     # Function to handle query processing
     def process_query_chat(self, query):
@@ -404,7 +428,7 @@ def process_query_chat(self, query):
             self.query_type, apply_llm_before_rag=apply_llm_before_rag
         )
 
-        if self.query_type == "Dataset":
+        if self.query_type == "Dataset" or self.query_type == "Flow":
             if config["structured_query"]:
                 # get structured query
                 response_parser.fetch_structured_query(self.query_type, query)
@@ -439,3 +463,6 @@ def process_query_chat(self, query):
 
             results = response_parser.parse_and_update_response(self.data_metadata)
             return results
+        elif self.query_type == "General Query":
+            response_parser.fetch_documentation_query(query)
+            return response_parser.documentation_response

pyproject.toml

@@ -50,6 +50,7 @@ urllib3 = "1.26.19"
 uvicorn = "0.29.0"
 poetry = "1.8.3"
 protobuf = "3.20.0"
+langchain-ollama = "0.1.1"
 
 
 [build-system]

start_local.sh

@@ -24,6 +24,10 @@ else
     echo $! > $PID_FILE
 fi
 
+cd ../documentation_bot || exit
+uvicorn documentation_query:app --host 0.0.0.0 --port 8083 &
+echo $! >> $PID_FILE
+
 cd ../backend || exit
 uvicorn backend:app --host 0.0.0.0 --port 8000 &
 echo $! >> $PID_FILE

stop_local.sh

@@ -20,6 +20,9 @@ cd ../frontend
 # streamlit run ui.py &
 kill -9 $(cat $PID_FILE)
 
+cd ../documentation_bot
+kill -9 $(cat $PID_FILE)
+
 cd ..
 
 killall ollama