Update app.py

app.py

@@ -4,56 +4,54 @@ from langchain_community.document_loaders import PyPDFLoader
 from langchain_text_splitters import RecursiveCharacterTextSplitter
 from langchain_community.embeddings import HuggingFaceEmbeddings
 from langchain_community.vectorstores import FAISS
-
 from langchain_openai import ChatOpenAI
-from langchain_core.prompts import ChatPromptTemplate
-from langchain.chains.combine_documents import create_stuff_documents_chain
-from langchain.chains import create_retrieval_chain
 
 
-def build_chain(pdf_path: str):
+def run_qa(pdf_path, question):
+    if pdf_path is None or not question or question.strip() == "":
+        return "Please upload a PDF and enter a question."
+
+    # 1) Load PDF
     loader = PyPDFLoader(pdf_path)
     docs = loader.load()
 
+    # 2) Split
     splitter = RecursiveCharacterTextSplitter(chunk_size=1000, chunk_overlap=150)
     chunks = splitter.split_documents(docs)
 
+    # 3) Embed + Vector store
     embeddings = HuggingFaceEmbeddings(model_name="all-MiniLM-L6-v2")
     vectordb = FAISS.from_documents(chunks, embeddings)
+
+    # 4) Retrieve relevant chunks
     retriever = vectordb.as_retriever(search_kwargs={"k": 4})
+    retrieved_docs = retriever.get_relevant_documents(question)
+
+    context = "\n\n".join([d.page_content for d in retrieved_docs])
 
+    # 5) LLM (OpenAI)
     llm = ChatOpenAI(temperature=0)
 
-    prompt = ChatPromptTemplate.from_template(
-        """You are a helpful assistant. Answer the question using ONLY the provided context.
-If the answer is not in the context, say you don't know.
+    prompt = f"""
+You are a helpful assistant. Answer the question using ONLY the context below.
+If the answer is not in the context, say "I don't know".
 
-<context>
+CONTEXT:
 {context}
-</context>
 
-Question: {input}
-"""
-    )
-
-    doc_chain = create_stuff_documents_chain(llm, prompt)
-    retrieval_chain = create_retrieval_chain(retriever, doc_chain)
-    return retrieval_chain
-
-
-def run_qa(pdf_path, question):
-    if pdf_path is None or question.strip() == "":
-        return "Please upload a PDF and enter a question."
+QUESTION:
+{question}
 
-    chain = build_chain(pdf_path)
-    result = chain.invoke({"input": question})
+Answer:
+""".strip()
 
-    answer_text = result.get("answer", "")
-    ctx_docs = result.get("context", []) or []
+    response = llm.invoke(prompt)
+    answer = response.content if hasattr(response, "content") else str(response)
 
-    sources = "\n\n".join([d.page_content[:500] for d in ctx_docs[:2]])
+    # 6) Sources preview
+    sources = "\n\n".join([d.page_content[:500] for d in retrieved_docs[:2]])
 
-    return f"### Answer\n{answer_text}\n\n---\n### Sources\n{sources}"
+    return f"### Answer\n{answer}\n\n---\n### Sources\n{sources}"
 
 
 with gr.Blocks(title="Agentic Document Intelligence") as demo: