📚 Build a RAG Chatbot with LangChain and Gemini

LangChain can load PDFs, text files, and web pages. Split them into overlapping chunks.

from langchain_community.document_loaders import PyPDFLoader, TextLoader
from langchain.text_splitter import RecursiveCharacterTextSplitter

# Load a PDF (swap for TextLoader("notes.txt") for plain text)
loader = PyPDFLoader("handbook.pdf")
docs   = loader.load()

splitter = RecursiveCharacterTextSplitter(chunk_size=1000, chunk_overlap=200)
chunks   = splitter.split_documents(docs)
print(f"Split into {len(chunks)} chunks")

Split into 84 chunks

from langchain_community.embeddings import HuggingFaceEmbeddings
from langchain_community.vectorstores import Chroma

embeddings = HuggingFaceEmbeddings(model_name="all-MiniLM-L6-v2")
vectorstore = Chroma.from_documents(chunks, embeddings, persist_directory="./chroma_db")
print("Vector store built and saved.")

Vector store built and saved.

from langchain_google_genai import ChatGoogleGenerativeAI
from langchain.chains import RetrievalQA
import os

os.environ["GOOGLE_API_KEY"] = "YOUR_GEMINI_API_KEY"

llm = ChatGoogleGenerativeAI(model="gemini-1.5-flash", temperature=0.2)
retriever = vectorstore.as_retriever(search_kwargs={"k": 4})

qa_chain = RetrievalQA.from_chain_type(
    llm=llm,
    chain_type="stuff",
    retriever=retriever,
    return_source_documents=True,
)

questions = [
    "What is the refund policy?",
    "How do I reset my password?",
]

for q in questions:
    result = qa_chain({"query": q})
    print(f"Q: {q}")
    print(f"A: {result['result']}\n")
    print("Sources:")
    for doc in result["source_documents"]:
        print(f"  - Page {doc.metadata.get('page','?')+1}: {doc.page_content[:80]}...")
    print()

Q: What is the refund policy?
A: Refunds are available within 30 days of purchase for unused accounts.

Sources:
  - Page 12: Refunds and Cancellations. All refund requests must be submitted within...