LangChain ke Saath PDF Chatbot

🔧 Prerequisite

Sabse pehle humein kuch libraries install karni hongi:

pip install langchain pypdf python-dotenv openai chromadb

Step 1: PDF Loader Setup 📚

PDF ko load karne ke liye hum PyPDFLoader ka use karenge:

from langchain_community.document_loaders import PyPDFLoader

# PDF file path
file_path = "your_document.pdf"

# PDF loader initialize karna
loader = PyPDFLoader(file_path)

# Documents load karna
documents = loader.load()

Explanation:

• PyPDFLoader PDF file ko load karta hai.
• .load() method saare pages ko document objects me convert karta hai.
• Har page ek separate document ban jata hai.

Step 2: Text Splitting ✂️

PDF ke text ko chunks me split karna important hai:

from langchain.text_splitter import RecursiveCharacterTextSplitter

# Text splitter create karna
text_splitter = RecursiveCharacterTextSplitter(
    chunk_size=1000,
    chunk_overlap=200
)

# Documents ko split karna
texts = text_splitter.split_documents(documents)

Explanation:

• chunk_size=1000 → har chunk 1000 characters ka hoga.
• chunk_overlap=200 → 200 characters overlap rahenge taaki context na toote.
• Yeh approach semantic meaning ko preserve karta hai.

Step 3: Embeddings Generate Karna 🧠

Embeddings document ke meaning ko vector format me convert karte hain:

from langchain_openai import OpenAIEmbeddings
from langchain_community.vectorstores import Chroma

# OpenAI Embeddings
embeddings = OpenAIEmbeddings()

# Vector Store create karna
vectorstore = Chroma.from_documents(
    documents=texts,
    embedding=embeddings
)

Explanation:

• OpenAIEmbeddings text ko high-dimensional vector me convert karta hai.
• Chroma vector store documents ko efficiently store & retrieve karne me help karta hai.

Step 4: Retrieval Chain Setup 🔗

Retrieval chain document se relevant information nikalne me help karta hai:

from langchain.chains import RetrievalQA
from langchain_openai import ChatOpenAI

# ChatGPT model initialize
llm = ChatOpenAI(model_name="gpt-3.5-turbo")

# Retrieval QA Chain
qa_chain = RetrievalQA.from_chain_type(
    llm=llm,
    chain_type="stuff",
    retriever=vectorstore.as_retriever()
)

Explanation:

• ChatOpenAI language model ko initialize karta hai.
• RetrievalQA chain documents se relevant context nikalta hai.
• chain_type="stuff" → context directly model ko pass hota hai.

Step 5: Chatbot Interaction 💬

Ab aap apne PDF se interact kar sakte hain!

# Query karna
query = "Aapke document ka main point kya hai?"
result = qa_chain.invoke(query)

print(result['result'])

Yeh command aapke PDF ke content se relevant jawab nikal kar dikhata hai 🎯

Advanced Features 🚀

1. Conversation Memory

from langchain.memory import ConversationBufferMemory
from langchain.chains import ConversationalRetrievalChain

memory = ConversationBufferMemory()

Memory feature chatbot ko past questions ya context yaad rakhne me help karta hai — taaki conversation natural lage aur user experience better ho.