Update app.py

app.py

@@ -1,6 +1,5 @@
 # app.py
 
-import os
 import logging
 import re
 import requests
@@ -17,44 +16,37 @@ from langchain.chains.summarize import load_summarize_chain
 from langchain.docstore.document import Document
 from langchain.chains import RetrievalQA
 
-# Load your Together API key securely (recommended on HF Spaces)
-TOGETHER_API_KEY = os.environ.get("TOGETHER_API_KEY")
-
 # Logging setup
 logging.basicConfig(level=logging.INFO)
 logger = logging.getLogger(__name__)
 
-# Load models
-logger.info("🔍 Loading sentence transformer and LLM...")
+# Load Embedding Model
+logger.info("🔍 Loading sentence transformer...")
 embed_model = SentenceTransformer("all-MiniLM-L6-v2")
+
+# Load LLM (Replace with your API Key)
 llm = Together(
     model="togethercomputer/llama-3-70b-chat",
     temperature=0.7,
     max_tokens=512,
-    together_api_key=TOGETHER_API_KEY,
+    together_api_key="your_together_api_key"
 )
 
-# Global cache
-vector_index = None
-doc_chunks = []
-doc_texts = []
-doc_embeddings = []
-
-# Helper Functions
 def fetch_webpage_text(url):
     try:
         response = requests.get(url)
         response.raise_for_status()
         soup = BeautifulSoup(response.text, "html.parser")
-        content = soup.find("div", {"id": "mw-content-text"}) or soup.body
-        return content.get_text(separator="\n", strip=True)
+        content_div = soup.find("div", {"id": "mw-content-text"}) or soup.body
+        return content_div.get_text(separator="\n", strip=True)
     except Exception as e:
-        logger.error(f"❌ Error fetching content: {e}")
+        logger.error(f"Error fetching content: {e}")
         return ""
 
 def clean_text(text):
     text = re.sub(r'\[\s*\d+\s*\]', '', text)
     text = re.sub(r'\[\s*[a-zA-Z]+\s*\]', '', text)
+    text = re.sub(r'^\[\s*\d+\s*\]$', '', text, flags=re.MULTILINE)
     text = re.sub(r'\n{2,}', '\n', text)
     text = re.sub(r'[ \t]+', ' ', text)
     return text.strip()
@@ -79,63 +71,49 @@ def get_summary(chunks):
     summarize_chain = load_summarize_chain(llm, chain_type="map_reduce")
     return summarize_chain.run([full_doc])
 
-def chat_with_bot(question):
-    if not doc_chunks or not doc_embeddings:
-        return "⚠️ Please load a webpage and summarize it first."
-
-    query_vector = embed_model.encode(question).astype(np.float32)
-    index = faiss.IndexFlatL2(doc_embeddings[0].shape[0])
-    index.add(np.array(doc_embeddings).astype(np.float32))
+def retrieve_answer(query, chunks, embeddings, texts):
+    query_vector = embed_model.encode(query).astype(np.float32)
+    index = faiss.IndexFlatL2(embeddings[0].shape[0])
+    index.add(np.array(embeddings).astype(np.float32))
     D, I = index.search(np.array([query_vector]), k=5)
-    top_chunks = [doc_texts[i] for i in I[0]]
-    rag_doc = "\n\n".join(top_chunks)
+    top_chunks = [texts[i] for i in I[0]]
 
+    rag_doc = "\n\n".join(top_chunks)
     qa_chain = RetrievalQA.from_chain_type(llm=llm, chain_type="stuff", retriever=None)
-    return qa_chain.run(input_documents=[Document(page_content=rag_doc)], question=question)
-
-def summarize_content():
-    if not doc_chunks:
-        return "⚠️ No content loaded yet. Please load a valid webpage."
-    return get_summary(doc_chunks)
-
-def process_webpage_and_load(url):
-    global doc_chunks, vector_index, doc_texts, doc_embeddings
-    logger.info(f"🌐 Loading URL: {url}")
-    text = fetch_webpage_text(url)
-    if not text:
-        return "❌ Failed to load or parse webpage."
-    cleaned = clean_text(text)
-    doc_chunks = chunk_text(cleaned)
-    vector_index, doc_texts, doc_embeddings = create_vectorstore(doc_chunks)
-    return "✅ Webpage content processed and ready!"
-
-# Gradio UI
-with gr.Blocks() as demo:
-    gr.Markdown("## 🤖 Chat with LLaMA Webpage Content")
-
-    with gr.Row():
-        chatbot = gr.Chatbot(label="Chat History")
-
-    with gr.Row():
-        question = gr.Textbox(
-            label="Ask your question about LLaMA",
-            placeholder="e.g., Who developed LLaMA?"
-        )
-        ask_btn = gr.Button("Submit")
-        clear_btn = gr.Button("Clear Chat")
-
-    summary_output = gr.Textbox(label="📋 Summary of the Webpage", lines=8)
-    summarize_btn = gr.Button("Summarize Content")
-
-    # Button logic
-    def user_chat_handler(q, history):
-        response = chat_with_bot(q)
-        history.append((q, response))
-        return history, ""
-
-    ask_btn.click(fn=user_chat_handler, inputs=[question, chatbot], outputs=[chatbot, question])
-    clear_btn.click(lambda: [], None, chatbot)
-    summarize_btn.click(fn=summarize_content, inputs=[], outputs=summary_output)
-
-demo.launch()
+    return qa_chain.run(input_documents=[Document(page_content=rag_doc)], question=query)
+
+# Gradio Interface
+def run_chatbot(url, query):
+    raw_text = fetch_webpage_text(url)
+    if not raw_text:
+        return "❌ Failed to fetch content.", ""
+
+    cleaned = clean_text(raw_text)
+    chunks = chunk_text(cleaned)
+
+    if not chunks:
+        return "❌ No valid content to process.", ""
+
+    summary = get_summary(chunks)
+    index, texts, embeddings = create_vectorstore(chunks)
+    answer = retrieve_answer(query, chunks, embeddings, texts)
+
+    return summary, answer
+
+demo = gr.Interface(
+    fn=run_chatbot,
+    inputs=[
+        gr.Textbox(label="Webpage URL", placeholder="Enter a Wikipedia link"),
+        gr.Textbox(label="Your Question", placeholder="Ask a question about the webpage")
+    ],
+    outputs=[
+        gr.Textbox(label="Webpage Summary"),
+        gr.Textbox(label="Answer")
+    ],
+    title="🦙 LLaMA RAG Chatbot",
+    description="Enter a Wikipedia article URL and ask a question. Powered by Together AI and LangChain.",
+    allow_flagging="never"
+)
 
+if __name__ == "__main__":
+    demo.launch()