Simple search engine gui development

web_gui/streamlit.py

@@ -8,6 +8,9 @@ from sentence_transformers import SentenceTransformer
 from sklearn.metrics.pairwise import cosine_similarity
 from huggingface_hub import InferenceClient
 
+###### Import search engine from directory
+from simple_search_engine.search_engine import SimpleSearchEngine
+
 # --- Configuration & Constants ---
 PAGE_TITLE = "Manga & TV Assistant"
 DATA_DIRECTORY = "corpus"
@@ -27,6 +30,17 @@ SYSTEM_PROMPT = HYDE_SYSTEM_PROMPT =(
 
 st.set_page_config(page_title=PAGE_TITLE, layout="wide")
 
+###### Added - Radio button to switch between simple search engine and RAG agent
+def render_mode_selector():
+    with st.sidebar:
+        st.header("Mode")
+        mode = st.radio(
+            "Select application mode",
+            ["RAG Chat", "TF-IDF Search"],
+            key="app_mode")
+        st.divider()
+    return mode
+#####
 
 # --- Helper Functions: Data Loading & Processing ---
 
@@ -398,45 +412,85 @@ def render_chat_interface():
 
 
 # --- Main Application Execution ---
-
-initialize_session_state()
-
-if st.session_state['doc_embeddings'] is None:
-    process_documents_and_embed()
-
-render_sidebar()
-msg_container = render_chat_interface()
-
-if prompt := st.chat_input("Ask a question..."):
-    msg_container.chat_message("user", avatar=":material/psychology_alt:").markdown(prompt)
-
-    col1, col2 = msg_container.columns(2)
-
-    with col1:
-        with st.spinner("Standard RAG..."):
-            sim_results = find_similar_context(prompt)
-            if sim_results['max_similarity'] > st.session_state['similarity_threshold']:
-                std_response, _ = generate_rag_response(prompt, sim_results['sentence_indices'], sim_results['sources'])
-            else:
-                std_response = f"Low similarity ({sim_results['max_similarity']:.2f}). No relevant info found."
-            st.markdown("### Standard RAG")
-            st.markdown(std_response)
-
-    with col2:
-        with st.spinner("HyDE processing..."):
-            hyde_response, hyde_hypothetical, hyde_sim_results, hyde_score = run_hyde_process(prompt)
-            st.markdown("### HyDE Response")
-            st.markdown(hyde_response)
-
-    # Save history
-    st.session_state['chat_history'].append({"role": "user", "content": prompt})
-    st.session_state['chat_history'].append({"role": "assistant", "content": std_response, "type": "normal"})
-    st.session_state['chat_history'].append({"role": "assistant", "content": hyde_response, "type": "hyde"})
-
-    # Update Session State for Sidebar
-    st.session_state['last_std_sources'] = sim_results.get('sources', set())
-    st.session_state['last_hyde_sources'] = hyde_sim_results.get('sources', set())
-    st.session_state['last_hyde_hypothetical'] = hyde_hypothetical
-
-    # Rerun to update sidebar immediately
-    st.rerun()
+###### Replaced below and indented once to the right - Mode check
+#initialize_session_state()
+
+#if st.session_state['doc_embeddings'] is None:
+#    process_documents_and_embed()
+
+#render_sidebar()
+#msg_container = render_chat_interface()
+######
+
+mode = render_mode_selector()
+
+if mode == "RAG Chat":
+    initialize_session_state()
+
+    if st.session_state['doc_embeddings'] is None:
+        process_documents_and_embed()
+
+    render_sidebar()
+    msg_container = render_chat_interface()
+
+    if prompt := st.chat_input("Ask a question..."):
+        msg_container.chat_message("user", avatar=":material/psychology_alt:").markdown(prompt)
+    
+        col1, col2 = msg_container.columns(2)
+    
+        with col1:
+            with st.spinner("Standard RAG..."):
+                sim_results = find_similar_context(prompt)
+                if sim_results['max_similarity'] > st.session_state['similarity_threshold']:
+                    std_response, _ = generate_rag_response(prompt, sim_results['sentence_indices'], sim_results['sources'])
+                else:
+                    std_response = f"Low similarity ({sim_results['max_similarity']:.2f}). No relevant info found."
+                st.markdown("### Standard RAG")
+                st.markdown(std_response)
+    
+        with col2:
+            with st.spinner("HyDE processing..."):
+                hyde_response, hyde_hypothetical, hyde_sim_results, hyde_score = run_hyde_process(prompt)
+                st.markdown("### HyDE Response")
+                st.markdown(hyde_response)
+    
+        # Save history
+        st.session_state['chat_history'].append({"role": "user", "content": prompt})
+        st.session_state['chat_history'].append({"role": "assistant", "content": std_response, "type": "normal"})
+        st.session_state['chat_history'].append({"role": "assistant", "content": hyde_response, "type": "hyde"})
+    
+        # Update Session State for Sidebar
+        st.session_state['last_std_sources'] = sim_results.get('sources', set())
+        st.session_state['last_hyde_sources'] = hyde_sim_results.get('sources', set())
+        st.session_state['last_hyde_hypothetical'] = hyde_hypothetical
+    
+        # Rerun to update sidebar immediately
+        st.rerun()
+        
+###### Added TF-IDF logic
+if mode == "TF-IDF Search":
+    if "tfidf_engine" not in st.session_state:
+        engine = SimpleSearchEngine(corpus_dir="corpus")
+        engine.build_index()
+        st.session_state["tfidf_engine"] = engine
+    st.title("TF-IDF Search Engine")
+    
+    query = st.text_input("Search",placeholder="Search the One Piece corpus like Google…")
+    
+    if query:
+        results = st.session_state["tfidf_engine"].search(query, top_k=10)
+    
+        if not results:
+            st.info("No results found.")
+        else:
+            st.caption(f"Showing {len(results)} results")
+    
+            for i, r in enumerate(results, start=1):
+                st.markdown(f"### {i}. {r['title']}")
+                st.caption(f"Relevance score: {r['score']:.4f}")
+    
+                if r.get("url"):
+                    st.markdown(r["url"])
+    
+                st.write(r["snippet"])
+                st.divider()