Update app.py

app.py

@@ -1,173 +1,185 @@
-import streamlit as st
-from generator import generate_response_from_document
-from retrieval import retrieve_documents_hybrid,find_query_dataset
-from evaluation import calculate_metrics
-from data_processing import load_recent_questions, save_recent_question
-import time
-import matplotlib.pyplot as plt
-
-# Page Title
-st.title("RAG7 - Real World RAG System")
-
-st.markdown(
-    """
-    <style>
-    .stTextArea textarea {
-        background-color: white !important;
-        font-size: 20px !important;
-        color: black !important;
-    }
-    </style>
-    """,
-    unsafe_allow_html=True
-)
-
-# global retrieved_documents
-# retrieved_documents = []
-
-# global response 
-# response = ""
-
-# global time_taken_for_response
-# time_taken_for_response = 'N/A'
-
-# @st.cache_data
-# def load_data():
-#     load_data_from_faiss()
-
-# data_status = load_data()
-
-# Question Section
-st.subheader("Hi, What do you want to know today?")
-question = st.text_area("Enter your question:", placeholder="Type your question here...", height=100)
-question = question.strip()
-
-# # Submit Button
-# if st.button("Submit"):
-#     start_time = time.time()
-#     retrieved_documents = retrieve_documents_hybrid(question, 10)  
-#     response = generate_response_from_document(question, retrieved_documents)
-#     end_time = time.time()
-#     time_taken_for_response = end_time-start_time
-# else:
-#     response = ""
-
-# # Response Section
-# st.subheader("Response")
-# st.text_area("Generated Response:", value=response, height=150, disabled=True)
-
-# # Metrics Section
-# st.subheader("Metrics")
-
-# col1, col2 = st.columns([1, 3])  # Creating two columns for button and metrics display
-
-# with col1:
-#     if st.button("Calculate Metrics"):
-#         metrics = calculate_metrics(question, response, retrieved_documents, time_taken_for_response)
-#     else:
-#         metrics = ""
-
-# with col2:
-#     st.text_area("Metrics:", value=metrics, height=100, disabled=True)
-
-if "retrieved_documents" not in st.session_state:
-    st.session_state.retrieved_documents = []
-if "response" not in st.session_state:
-    st.session_state.response = ""
-if "time_taken_for_response" not in st.session_state:
-    st.session_state.time_taken_for_response = "N/A"
-if "metrics" not in st.session_state:
-    st.session_state.metrics = {}
-if "query_dataset" not in st.session_state:
-    st.session_state.query_dataset = ''
-
-recent_questions = load_recent_questions()
-
-# for visualization
-
-# response_time = [q["response_time"] for q in recent_data["questions"]]
-# labels = [f"Q{i+1}" for i in range(len(response_time))]  # Labels for X-axis
-
-# fig, ax = plt.subplots()
-# ax.set_xlabel("Recent Questions")
-# ax.set_ylabel("Time Taken for Response")
-# ax.legend()
-# st.sidebar.pyplot(fig)
-if recent_questions and "questions" in recent_questions and recent_questions["questions"]:
-    recent_qns = list(reversed(recent_questions["questions"]))
-    st.sidebar.title("Analytics")
-
-    # Extract response times and labels
-    response_time = [q["response_time"] for q in recent_qns]
-    labels = [f"Q{i+1}" for i in range(len(response_time))]
-
-    # Plot graph
-    fig, ax = plt.subplots()
-    ax.plot(labels, response_time, marker="o", linestyle="-", color="skyblue")
-    ax.set_xlabel("Recent Questions")
-    ax.set_ylabel("Time Taken for Response (seconds)")
-    ax.set_title("Response Time Analysis")
-
-    # Display the plot in the sidebar
-    st.sidebar.pyplot(fig)
-
-    st.sidebar.markdown("---")
-
-    # Display Recent Questions
-    st.sidebar.title("Recent Questions")
-    for q in recent_qns:  # Show latest first
-        st.sidebar.write(f"🔹 {q['question']}")
-else:
-    st.sidebar.write("No recent questions")
- # Separator
-
-# Streamlit Sidebar for Recent Questions
-
-
-# Submit Button
-# if st.button("Submit"):
-#     start_time = time.time()
-#     st.session_state.retrieved_documents = retrieve_documents_hybrid(question, 10)  
-#     st.session_state.response = generate_response_from_document(question, st.session_state.retrieved_documents)
-#     end_time = time.time()
-#     st.session_state.time_taken_for_response = end_time - start_time
-
-if st.button("Submit"):
-    start_time = time.time()
-    st.session_state.metrics = {}
-    st.session_state.query_dataset =  find_query_dataset(question)
-    st.session_state.retrieved_documents = retrieve_documents_hybrid(question, st.session_state.query_dataset, 10)  
-    st.session_state.response = generate_response_from_document(question, st.session_state.retrieved_documents)
-    end_time = time.time()
-    st.session_state.time_taken_for_response = end_time - start_time
-    save_recent_question(question, st.session_state.time_taken_for_response)
-
-# Display stored response
-st.subheader("Response")
-st.text_area("Generated Response:", value=st.session_state.response, height=150, disabled=True)
-
-col1, col2 = st.columns([1, 3])  # Creating two columns for button and metrics display
-
-# # Calculate Metrics Button
-# with col1:
-#     if st.button("Calculate Metrics"):
-#         metrics = calculate_metrics(question, st.session_state.response, st.session_state.retrieved_documents, st.session_state.time_taken_for_response)
-#     else:
-#         metrics = {}
-
-# with col2:
-#     #st.text_area("Metrics:", value=metrics, height=100, disabled=True)
-#     st.json(metrics)
-
-
-# Calculate Metrics Button
-with col1:
-    if st.button("Show Metrics"):     
-        st.session_state.metrics = calculate_metrics(question, st.session_state.query_dataset, st.session_state.response, st.session_state.retrieved_documents, st.session_state.time_taken_for_response)
-    else:
-        metrics_ = {}
-
-with col2:
-    #st.text_area("Metrics:", value=metrics, height=100, disabled=True)
-    st.json(st.session_state.metrics)
-
+import streamlit as st
+from generator import generate_response_from_document
+from retrieval import retrieve_documents_hybrid,find_query_dataset
+from evaluation import calculate_metrics
+from data_processing import load_recent_questions, save_recent_question
+import time
+import matplotlib.pyplot as plt
+
+# Page Title
+st.title("RAG7 - Real World RAG System")
+
+st.markdown(
+    """
+    <style>
+    .stTextArea textarea {
+        background-color: white !important;
+        font-size: 20px !important;
+        color: black !important;
+    }
+    </style>
+    """,
+    unsafe_allow_html=True
+)
+
+# global retrieved_documents
+# retrieved_documents = []
+
+# global response 
+# response = ""
+
+# global time_taken_for_response
+# time_taken_for_response = 'N/A'
+
+# @st.cache_data
+# def load_data():
+#     load_data_from_faiss()
+
+# data_status = load_data()
+
+# Question Section
+st.subheader("Hi, What do you want to know today?")
+question = st.text_area("Enter your question:", placeholder="Type your question here...", height=100)
+question = question.strip()
+
+# # Submit Button
+# if st.button("Submit"):
+#     start_time = time.time()
+#     retrieved_documents = retrieve_documents_hybrid(question, 10)  
+#     response = generate_response_from_document(question, retrieved_documents)
+#     end_time = time.time()
+#     time_taken_for_response = end_time-start_time
+# else:
+#     response = ""
+
+# # Response Section
+# st.subheader("Response")
+# st.text_area("Generated Response:", value=response, height=150, disabled=True)
+
+# # Metrics Section
+# st.subheader("Metrics")
+
+# col1, col2 = st.columns([1, 3])  # Creating two columns for button and metrics display
+
+# with col1:
+#     if st.button("Calculate Metrics"):
+#         metrics = calculate_metrics(question, response, retrieved_documents, time_taken_for_response)
+#     else:
+#         metrics = ""
+
+# with col2:
+#     st.text_area("Metrics:", value=metrics, height=100, disabled=True)
+
+if "retrieved_documents" not in st.session_state:
+    st.session_state.retrieved_documents = []
+if "response" not in st.session_state:
+    st.session_state.response = ""
+if "time_taken_for_response" not in st.session_state:
+    st.session_state.time_taken_for_response = "N/A"
+if "metrics" not in st.session_state:
+    st.session_state.metrics = {}
+if "query_dataset" not in st.session_state:
+    st.session_state.query_dataset = ''
+
+recent_questions = load_recent_questions()
+
+# for visualization
+
+# response_time = [q["response_time"] for q in recent_data["questions"]]
+# labels = [f"Q{i+1}" for i in range(len(response_time))]  # Labels for X-axis
+
+# fig, ax = plt.subplots()
+# ax.set_xlabel("Recent Questions")
+# ax.set_ylabel("Time Taken for Response")
+# ax.legend()
+# st.sidebar.pyplot(fig)
+if recent_questions and "questions" in recent_questions and recent_questions["questions"]:
+    recent_qns = list(reversed(recent_questions["questions"]))
+    st.sidebar.title("Analytics")
+
+    # Extract response times and labels
+    response_time = [q["response_time"] for q in recent_qns]
+    labels = [f"Q{i+1}" for i in range(len(response_time))]
+
+    # Plot graph
+    fig, ax = plt.subplots()
+    ax.plot(labels, response_time, marker="o", linestyle="-", color="skyblue")
+    ax.set_xlabel("Recent Questions")
+    ax.set_ylabel("Time Taken for Response (seconds)")
+    ax.set_title("Response Time Analysis")
+
+    # Display the plot in the sidebar
+    st.sidebar.pyplot(fig)
+
+    st.sidebar.markdown("---")
+
+    # Display Recent Questions
+    st.sidebar.title("Recent Questions")
+    for q in recent_qns:  # Show latest first
+        st.sidebar.write(f"🔹 {q['question']}")
+else:
+    st.sidebar.write("No recent questions")
+ # Separator
+
+# Streamlit Sidebar for Recent Questions
+
+
+# Submit Button
+# if st.button("Submit"):
+#     start_time = time.time()
+#     st.session_state.retrieved_documents = retrieve_documents_hybrid(question, 10)  
+#     st.session_state.response = generate_response_from_document(question, st.session_state.retrieved_documents)
+#     end_time = time.time()
+#     st.session_state.time_taken_for_response = end_time - start_time
+
+if st.button("Submit"):
+    start_time = time.time()
+    st.session_state.metrics = {}
+    st.session_state.query_dataset =  find_query_dataset(question)
+    st.session_state.retrieved_documents = retrieve_documents_hybrid(question, st.session_state.query_dataset, 10)  
+    st.session_state.response = generate_response_from_document(question, st.session_state.retrieved_documents)
+    end_time = time.time()
+    st.session_state.time_taken_for_response = end_time - start_time
+    save_recent_question(question, st.session_state.time_taken_for_response)
+
+# Display stored response
+st.subheader("Response")
+st.text_area("Generated Response:", value=st.session_state.response, height=150, disabled=True)
+
+col1, col2 = st.columns([1, 3])  # Creating two columns for button and metrics display
+
+# # Calculate Metrics Button
+# with col1:
+#     if st.button("Calculate Metrics"):
+#         metrics = calculate_metrics(question, st.session_state.response, st.session_state.retrieved_documents, st.session_state.time_taken_for_response)
+#     else:
+#         metrics = {}
+
+# with col2:
+#     #st.text_area("Metrics:", value=metrics, height=100, disabled=True)
+#     st.json(metrics)
+
+
+# Calculate Metrics Button
+with col1:
+    if st.button("Show Metrics"):     
+        st.session_state.metrics = calculate_metrics(question, st.session_state.query_dataset, st.session_state.response, st.session_state.retrieved_documents, st.session_state.time_taken_for_response)
+    else:
+        metrics_ = {}
+
+with col2:
+    #st.text_area("Metrics:", value=metrics, height=100, disabled=True)
+    st.json(st.session_state.metrics)
+    
+# Calculate RMSE Button
+with col1:
+    if st.button("Show RMSE"):     
+        st.session_state.rmse = compute_rmse( st.session_state.metrics.values(),st.session_state.response)
+    else:
+        metrics_ = {}
+
+with col2:
+    #st.text_area("RMSE:", value=rmse, height=100, disabled=True)
+    st.json(st.session_state.rmse)
+
+