Update app.py

app.py

@@ -61,14 +61,6 @@ avatar_12 = get_image_base64("avatar_12.png")
 icon_base64 = get_image_base64("clipboard.png")
 
 
-# Load your sentiment analysis data (replace 'your_data.csv' with your actual file)
-df = pd.read_csv('./data/Consumer_sentiment_analysis_results.csv')
-
-
-
-    
-
-
 config = configparser.ConfigParser()
 # Set page to wide mode
 st.set_page_config(layout="wide")
@@ -90,9 +82,14 @@ google_sheet_url = os.getenv("Google_Sheet")
 sheet = client.open_by_url(google_sheet_url)
 worksheet = sheet.get_worksheet(0)
 
+# Retrieve the API key from the environment variables
+api_key = os.getenv("OPENAI_API_KEY")
 
-aoc_qa = None
+# Check if the API key is available, if not, raise an error
+if api_key is None:
+    raise ValueError("API key not found. Ensure that the OPENAI_API_KEY environment variable is set.")
 
+aoc_qa = None
 
 # Function to create a copy-to-clipboard button
 def create_copy_button(text_to_copy):
@@ -125,16 +122,6 @@ def create_copy_button(text_to_copy):
     return copy_js
 
 
-
-# Retrieve the API key from the environment variables
-api_key = os.getenv("OPENAI_API_KEY")
-
-# Check if the API key is available, if not, raise an error
-if api_key is None:
-    raise ValueError("API key not found. Ensure that the OPENAI_API_KEY environment variable is set.")
-
-
-
 # Create a Chroma database instance using the selected directory
 def create_chroma_instance(directory):
     # Create and return a Chroma database instance
@@ -145,9 +132,7 @@ def create_chroma_instance(directory):
 vectordb = Chroma()
 
 
-
-
-# Define the system message template
+# Define the system message template (Prompt Template)
 system_template = """You are an AI assistant created by Citizens Information.
 Most important rule: You have no knowledge other than the below context.
 Only use the below context to answer questions. If you don't know the answer from the context, say that you don't know. 
@@ -178,19 +163,13 @@ messages = [
 qa_prompt = ChatPromptTemplate.from_messages(messages)
 
 
-
-
 # Define the K Value
 k_value = 6
 
 # Define the search_type
 selected_search_type = 'similarity'
 
-
-
-
 chat_history = []
-user_query = ""
 
 answer = ""  # Initialize ai_response with a default value 
 
@@ -202,10 +181,6 @@ def ask_alans_ai(query, vectordb, chat_history, aoc_qa):
     # Call pdf_qa with the filtered chat history
     result = aoc_qa.invoke({"question": query, "chat_history": filtered_chat_history, "vectordb": vectordb})
 
-    # Create the filenames string using the result
-    # filenames = "["+"][".join(set([x.metadata['filepath'] for x in result['source_documents']]))+"]"
-
-    
     answer = result["answer"]
 
     # Append the new query and its answer to the original chat history
@@ -216,18 +191,14 @@ def ask_alans_ai(query, vectordb, chat_history, aoc_qa):
 def clear_input_box():
     st.session_state["new_item"] = ""
 
-
-
 # Clean and prepare data for appending
 def clean_string(s):
     return s.replace("\n", " ").replace("\t", " ")
 
 
-   
-# Streamlit app
+###################### Streamlit app ####################################################
 def main():
 
-
     st.markdown(
         """
             <style>
@@ -247,9 +218,6 @@ def main():
     if 'selected_model' not in st.session_state:
         st.session_state['selected_model'] = 'gpt-3.5-turbo'
     
-
-
-    
     # Function to generate a unique session ID
     def generate_session_id():
         if 'session_id' not in st.session_state:
@@ -274,9 +242,7 @@ def main():
         </style>
         """, unsafe_allow_html=True)
 
-    
-  
-    # Sidebar
+    ######## Sidebar ##############
     st.sidebar.title("About Citizens Information Chatbot")
     st.sidebar.write("""**Health, Social Welfare, Employment, Money and Tax, Moving Country, Returning to Ireland, Housing, Education and Training, Travel and Recreation, Environment, Government in Ireland, Consumer, Death and Bereavement, Family and Relationships, Justice**
         <br><br>
@@ -299,29 +265,21 @@ def main():
         <br><br>
         **Using this chatbot means you accept these terms. For more detailed advice, consult the <a href="https://www.citizensinformation.ie/" target="_blank">Citizens Information Website</a>**""", unsafe_allow_html=True)
 
-
-
-    
-
-           
     
-    # Create a slider widget in the sidebar
+    # Create an AI Temp slider widget in the sidebar
     st.sidebar.header("Select AI Temperature:")
     ai_temp = st.sidebar.slider(label="Temperature", min_value=0.0, max_value=1.0, value=0.0, step=0.1)
 
      
-    
     # Streamlit slider for selecting the value of k
     st.sidebar.header("Select a K Value for Retrieval:")
     k_value = st.sidebar.slider('K Value', min_value=1, max_value=20, value=6)
 
- 
-    
-
     # Initialize the selected model in session state
     if 'selected_model' not in st.session_state:
         st.session_state.selected_model = 'gpt-3.5-turbo'
 
+    # Create an LLM dropdown select in the sidebar
     st.sidebar.header("Select Large Language Model")
     model_options = [
     'gpt-3.5-turbo', 
@@ -334,15 +292,10 @@ def main():
     st.session_state['selected_model'] = selected_model
 
 
-
-
-
-
     # Initialize the selected_directory in session state
     if 'selected_directory' not in st.session_state:
         st.session_state.selected_directory = './db_recursive_word'
 
-
     st.sidebar.header("Select Chroma Database")
     
     # Define the dropdown options and corresponding directories
@@ -354,22 +307,14 @@ def main():
     }
 
 
-    
     # Sidebar dropdown to select the database, with ChromaDB1 (./data) as the default
     selected_db = st.sidebar.selectbox("Select Chroma Database", db_options, index=0)  # Default to first model
-    # Display the selected value in a message
-    # st.write(f"Selection made: {selected_db}")
 
     # Get the corresponding directory for the selected option
     selected_directory = db_options[selected_db]
-    # Display the selected directory in a message
-    # st.write(f"corresponding directory: {selected_directory}")
-    
+
     # Initialize Chroma instance
     vectordb = create_chroma_instance(selected_directory)
-    # Display the selected directory in a message
-    # st.write(f"corresponding vector db: {vectordb}")
-
 
     # Initialize the selected search type in session state
     if 'selected_search_type' not in st.session_state:
@@ -388,24 +333,11 @@ def main():
     selected_search_type = search_type_options.get(selected_search_type, "similarity")
 
     
-    # Display the selected search typein a message
-    # st.write(f"Selection made: {selected_search_type}")
-
-
-
-
-
-    
-
-    
     # Display avatars side by side with selection buttons
     st.sidebar.header("Select an Avatar:")
 
     col1, col2, col3 = st.sidebar.columns(3)
 
-
-    
-
     # Initialize the selected avatar in session state
     if 'user_selected_avatar' not in st.session_state:
         st.session_state.user_selected_avatar = avatar_1
@@ -454,8 +386,7 @@ def main():
         if st.button("Select 12"):
             st.session_state.user_selected_avatar = avatar_12
 
-
-
+    ############ Set up the LangChain Conversational Retrieval Chain ################
     aoc_qa = ConversationalRetrievalChain.from_llm(
         ChatOpenAI(temperature=ai_temp, model_name=selected_model),
         retriever=vectordb.as_retriever(search_kwargs={'k': k_value}, search_type=selected_search_type),
@@ -464,11 +395,6 @@ def main():
         verbose=False, 
         combine_docs_chain_kwargs={"prompt": qa_prompt})  
 
-       
-  
- 
-
-
     # HTML for social media links with base64-encoded images
     social_media_html = f"""
         <p>Find us on social media:</p>
@@ -489,7 +415,6 @@ def main():
     # Add social media links to sidebar
     st.sidebar.markdown(social_media_html, unsafe_allow_html=True)
 
-    
 
     st.markdown("""
         <style>
@@ -532,9 +457,6 @@ def main():
         </style>
         """, unsafe_allow_html=True)
    
-   
-
-
     # Custom CSS to change the focus style of st.text_area
     custom_css = """
     <style>
@@ -549,8 +471,6 @@ def main():
     # Inject custom CSS with markdown
     st.markdown(custom_css, unsafe_allow_html=True)
 
-
-
     # Get the current date and time
     current_datetime = datetime.now()
 
@@ -664,7 +584,7 @@ def main():
                
                     
                     
-    # Container for chat messages
+    ############# Container for chat messages ##############
     with st.container():
         # Display chat history
         for question, answer in st.session_state.chat_history:
@@ -690,17 +610,15 @@ def main():
             st.empty()
 
 
-
     # Your combined string with the current date included
     combined_string = f"Question: {message}\n\nAnswer: {answer}\n\nDate: {date_string}\n\nhttps://www.citizensinformation.ie/"
     # Create a list with the three strings
     
-
     message_clean = clean_string(message)
     answer_clean = clean_string(answer)
     date_string_clean = clean_string(date_string)
 
-    # Check length (Google Sheets cells have a limit, typically 50000 characters)
+    # Check length 
     max_length = 50000
     message_clean = message_clean[:max_length]
     answer_clean = answer_clean[:max_length]
@@ -709,9 +627,6 @@ def main():
     # Append the cleaned data to the worksheet
     data_to_append = [message_clean, answer_clean, date_string, str(ai_temp), st.session_state['session_id'], st.session_state['selected_model'], str(k_value), selected_directory, selected_search_type]
 
-
-    
-    
    
     # Create and display the copy button only if answer has content
     if answer:            
@@ -721,16 +636,8 @@ def main():
         # Input fields to Google Sheet
         worksheet.append_row(data_to_append)
 
-        
-
+ 
 # Run the Streamlit app
 if __name__ == "__main__":
     main()
 
-# print("system_template is:", system_template, end="\n")
-# print("aoc_qa is:", aoc_qa, end="\n")
-# print("messages is:", messages, end="\n")
-# print("qa_prompt is:", qa_prompt, end="\n")
-
-
-