Update app.py

app.py

@@ -8,6 +8,7 @@ import os
 import time
 from langchain_groq import ChatGroq
 from langchain.chains.qa_with_sources.retrieval import RetrievalQAWithSourcesChain
+from langchain.prompts import PromptTemplate
 from bs4 import SoupStrainer
 
 # Load environment variables (optional)
@@ -77,15 +78,36 @@ st.markdown("""
     .stSpinner > div {
         color: #00ddeb;
     }
+    .footer {
+        display: flex;
+        align-items: center;
+        justify-content: center;
+        padding: 10px;
+        background: rgba(255, 255, 255, 0.1);
+        border-top: 1px solid rgba(255, 255, 255, 0.2);
+        position: fixed;
+        bottom: 0;
+        width: 100%;
+        color: #e0e0e0;
+        font-size: 14px;
+    }
+    .footer img {
+        margin-right: 10px;
+    }
     </style>
 """, unsafe_allow_html=True)
 
+# Display large logo at the top of the main page
+st.image("https://i.postimg.cc/2j0QWF3Z/Removal-575.png", width=300)
+
 # Set Streamlit app title
 st.title("WebChatter 💬")
 
 # Initialize session state
 if "index_created" not in st.session_state:
     st.session_state.index_created = False
+if "url_content" not in st.session_state:
+    st.session_state.url_content = None
 
 # Sidebar for URL input
 with st.sidebar:
@@ -99,9 +121,31 @@ main_container = st.container()
 # Initialize the Groq LLM
 llm = ChatGroq(
     api_key=GROQ_API_KEY,
-    model="llama3-70b-8192"
+    model="llama3-70b-8192",
+    max_tokens=2048  # Increased for detailed answers
 )
 
+# Custom prompt for detailed answers
+qa_prompt = PromptTemplate(
+    template="""You are an expert assistant tasked with providing detailed, extensive, and comprehensive answers. Use the provided context to answer the question thoroughly, including explanations, examples, and additional relevant information. If the context is limited, expand on the topic with your knowledge to ensure a complete response.
+
+Context: {context}
+
+Question: {question}
+
+Answer with sources: """
+)
+
+# Function to summarize URL content
+def summarize_content(content, llm):
+    summary_prompt = f"""Summarize the following content in 3-5 sentences, capturing the main points and key details:
+
+{content}
+
+Summary: """
+    summary = llm.invoke(summary_prompt).content
+    return summary
+
 def save_faiss_index(vectorstore, path):
     vectorstore.save_local(path)
 
@@ -117,7 +161,6 @@ if process_url_clicked:
             with st.spinner("Processing URL..."):
                 try:
                     st.text("Data Loading...Started...✅✅✅")
-                    # Use SoupStrainer to specify tags to parse
                     parse_only = SoupStrainer(['title', 'p', 'h1', 'h2', 'h3'])
                     loader = WebBaseLoader(
                         web_paths=[url.strip()],
@@ -130,6 +173,9 @@ if process_url_clicked:
                         st.error("No content loaded from URL. Try a different URL (e.g., https://www.bbc.com/news/science-environment-67299122).")
                         st.stop()
 
+                    # Store content for summarization
+                    st.session_state.url_content = "\n".join([doc.page_content for doc in data])
+
                     st.text("Text Splitter...Started...✅✅✅")
                     text_splitter = RecursiveCharacterTextSplitter(
                         separators=['\n\n', '\n', '.', ','],
@@ -154,6 +200,14 @@ if process_url_clicked:
                 except Exception as e:
                     st.error(f"Error processing URL: {str(e)}")
 
+# Display summary if content is available
+if st.session_state.url_content:
+    with main_container:
+        st.header("Summary of the URL Content")
+        with st.spinner("Generating summary..."):
+            summary = summarize_content(st.session_state.url_content, llm)
+            st.write(summary)
+
 # Query input with Ask button
 with main_container:
     st.header("Ask a Question")
@@ -169,7 +223,11 @@ if ask_clicked and query:
                 try:
                     embeddings = HuggingFaceEmbeddings(model_name="sentence-transformers/all-MiniLM-L6-v2")
                     vectorstore = load_faiss_index("faiss_index", embeddings)
-                    chain = RetrievalQAWithSourcesChain.from_llm(llm=llm, retriever=vectorstore.as_retriever())
+                    chain = RetrievalQAWithSourcesChain.from_llm(
+                        llm=llm,
+                        retriever=vectorstore.as_retriever(),
+                        question_prompt=qa_prompt
+                    )
                     result = chain({"question": query}, return_only_outputs=True)
 
                     if not result.get("answer"):
@@ -188,4 +246,15 @@ if ask_clicked and query:
                     else:
                         st.write("No sources found.")
                 except Exception as e:
-                    st.error(f"Error answering query: {str(e)}")
+                    st.error(f"Error answering query: {str(e)}")
+
+# Footer with tiny logo and text
+st.markdown(
+    """
+    <div class="footer">
+        <img src="https://i.postimg.cc/2j0QWF3Z/Removal-575.png" width="20">
+        WebChatter © 2025 | Developed by Mahatir Ahmed Tusher
+    </div>
+    """,
+    unsafe_allow_html=True
+)