Introduction to LangChain

pip install langchain langchain-openai langchain-community langchain-chroma

# script_id: day_038_langchain_basics/chat_models
from langchain_openai import ChatOpenAI
from langchain_core.messages import HumanMessage, SystemMessage

# Initialize the model
llm = ChatOpenAI(model="gpt-4o-mini", temperature=0.7)

# Simple invocation
response = llm.invoke([
    SystemMessage(content="You are a helpful assistant."),
    HumanMessage(content="What is Python?")
])

print(response.content)

# script_id: day_038_langchain_basics/prompt_templates
from langchain_core.prompts import ChatPromptTemplate, MessagesPlaceholder

# Simple template
simple_prompt = ChatPromptTemplate.from_template(
    "Explain {topic} in simple terms for a {audience}."
)

# Format the prompt
formatted = simple_prompt.format(topic="machine learning", audience="beginner")
print(formatted)

# Chat prompt with system message
chat_prompt = ChatPromptTemplate.from_messages([
    ("system", "You are a {role}. Be {tone}."),
    ("human", "{input}")
])

messages = chat_prompt.format_messages(
    role="teacher",
    tone="friendly",
    input="Explain recursion"
)

# script_id: day_038_langchain_basics/output_parsers
from langchain_core.output_parsers import StrOutputParser, JsonOutputParser
from pydantic import BaseModel, Field

# Simple string parser
parser = StrOutputParser()

# JSON parser with schema
class MovieReview(BaseModel):
    title: str = Field(description="Movie title")
    rating: int = Field(description="Rating from 1-10")
    summary: str = Field(description="Brief summary")

json_parser = JsonOutputParser(pydantic_object=MovieReview)

# Get format instructions for the prompt
print(json_parser.get_format_instructions())

# script_id: day_038_langchain_basics/lcel_basic_chain
from langchain_openai import ChatOpenAI
from langchain_core.prompts import ChatPromptTemplate
from langchain_core.output_parsers import StrOutputParser

# Define components
prompt = ChatPromptTemplate.from_template("Tell me a joke about {topic}")
model = ChatOpenAI(model="gpt-4o-mini")
parser = StrOutputParser()

# Compose the chain with LCEL
chain = prompt | model | parser

# Run the chain
result = chain.invoke({"topic": "programming"})
print(result)

# script_id: day_038_langchain_basics/chaining_multiple_steps
from langchain_openai import ChatOpenAI
from langchain_core.prompts import ChatPromptTemplate
from langchain_core.output_parsers import StrOutputParser

model = ChatOpenAI(model="gpt-4o-mini")

# Chain 1: Generate an outline
outline_prompt = ChatPromptTemplate.from_template(
    "Create a brief outline for an article about {topic}. Return just bullet points."
)

# Chain 2: Expand the outline
expand_prompt = ChatPromptTemplate.from_template(
    "Expand this outline into a short article:\n{outline}"
)

# Compose chains
outline_chain = outline_prompt | model | StrOutputParser()
expand_chain = expand_prompt | model | StrOutputParser()

# A dict in an LCEL pipe runs each value and builds a dict of results for the
# next stage — here it feeds the {outline} placeholder in expand_prompt.
full_chain = {"outline": outline_chain} | expand_chain

result = full_chain.invoke({"topic": "AI in healthcare"})
print(result)

# script_id: day_038_langchain_basics/document_loaders
from langchain_community.document_loaders import (
    TextLoader,
    PyPDFLoader,
    WebBaseLoader,
    DirectoryLoader
)

# Load a text file
text_loader = TextLoader("document.txt")
docs = text_loader.load()

# Load a PDF
pdf_loader = PyPDFLoader("report.pdf")
pdf_docs = pdf_loader.load()

# Load from web
web_loader = WebBaseLoader("https://example.com/article")
web_docs = web_loader.load()

# Load all files from directory
dir_loader = DirectoryLoader("./documents", glob="**/*.txt")
all_docs = dir_loader.load()

# Each document has content and metadata
for doc in docs:
    print(f"Content: {doc.page_content[:100]}...")
    print(f"Metadata: {doc.metadata}")

# script_id: day_038_langchain_basics/text_splitters
from langchain_text_splitters import (
    RecursiveCharacterTextSplitter,
    CharacterTextSplitter,
    TokenTextSplitter
)

# Recursive splitter (recommended)
splitter = RecursiveCharacterTextSplitter(
    chunk_size=1000,
    chunk_overlap=200,
    separators=["\n\n", "\n", ". ", " ", ""]
)

text = "Your long document text here..."
chunks = splitter.split_text(text)

# Or split documents directly
from langchain_community.document_loaders import TextLoader
loader = TextLoader("document.txt")
documents = loader.load()
split_docs = splitter.split_documents(documents)

print(f"Original: 1 document")
print(f"After split: {len(split_docs)} chunks")

# script_id: day_038_langchain_basics/rag_chain
from langchain_openai import ChatOpenAI, OpenAIEmbeddings
from langchain_chroma import Chroma
from langchain_core.prompts import ChatPromptTemplate
from langchain_core.runnables import RunnablePassthrough
from langchain_core.output_parsers import StrOutputParser

# Setup components
embeddings = OpenAIEmbeddings()
vectorstore = Chroma(embedding_function=embeddings, persist_directory="./db")
retriever = vectorstore.as_retriever(search_kwargs={"k": 3})
llm = ChatOpenAI(model="gpt-4o-mini")

# RAG prompt
rag_prompt = ChatPromptTemplate.from_template("""
Answer the question based on the following context:

Context: {context}

Question: {question}

Answer:""")

# Helper to format documents
def format_docs(docs):
    return "\n\n".join(doc.page_content for doc in docs)

# Build RAG chain with LCEL.
# The dict keys (context, question) must match the {placeholders} in the prompt.
# RunnablePassthrough() is a no-op stage that forwards the chain input unchanged —
# here it copies the incoming question into the prompt, while
# retriever | format_docs fetches and formats the matching documents.
rag_chain = (
    {"context": retriever | format_docs, "question": RunnablePassthrough()}
    | rag_prompt
    | llm
    | StrOutputParser()
)

# Use the chain (with error handling for production)
try:
    answer = rag_chain.invoke("What is machine learning?")
    print(answer)
except Exception as e:
    # LangChain wraps API errors — always catch broadly and log
    print(f"Chain failed: {type(e).__name__}: {e}")
    # In production: return a fallback response, log the error, alert if repeated

# script_id: day_038_langchain_basics/streaming_lcel
from langchain_openai import ChatOpenAI
from langchain_core.prompts import ChatPromptTemplate

prompt = ChatPromptTemplate.from_template("Write a story about {topic}")
model = ChatOpenAI(model="gpt-4o-mini", streaming=True)

chain = prompt | model

# Stream the response
for chunk in chain.stream({"topic": "a robot learning to paint"}):
    print(chunk.content, end="", flush=True)

# script_id: day_038_langchain_basics/async_support
import asyncio
from langchain_openai import ChatOpenAI
from langchain_core.prompts import ChatPromptTemplate

prompt = ChatPromptTemplate.from_template("Explain {topic} briefly")
model = ChatOpenAI(model="gpt-4o-mini")
chain = prompt | model

async def process_topics(topics: list):
    """Process multiple topics concurrently."""
    tasks = [chain.ainvoke({"topic": t}) for t in topics]
    results = await asyncio.gather(*tasks)
    return results

# Run async
topics = ["Python", "JavaScript", "Rust"]
results = asyncio.run(process_topics(topics))

for topic, result in zip(topics, results):
    print(f"{topic}: {result.content[:100]}...")

# script_id: day_038_langchain_basics/vanilla_vs_langchain
# Vanilla Python
from openai import OpenAI
client = OpenAI()

def vanilla_rag(question, docs):
    context = "\n".join(docs)
    response = client.chat.completions.create(
        model="gpt-4o-mini",
        messages=[
            {"role": "system", "content": f"Context: {context}"},
            {"role": "user", "content": question}
        ]
    )
    return response.choices[0].message.content

# LangChain
from langchain_openai import ChatOpenAI
from langchain_core.prompts import ChatPromptTemplate
from langchain_core.output_parsers import StrOutputParser

prompt = ChatPromptTemplate.from_template("Context: {context}\n\nQuestion: {question}")
chain = prompt | ChatOpenAI(model="gpt-4o-mini") | StrOutputParser()
result = chain.invoke({"context": docs, "question": question})

# script_id: day_038_langchain_basics/quick_reference
# Basic chain
from langchain_openai import ChatOpenAI
from langchain_core.prompts import ChatPromptTemplate
from langchain_core.output_parsers import StrOutputParser

chain = (
    ChatPromptTemplate.from_template("{input}")
    | ChatOpenAI()
    | StrOutputParser()
)
result = chain.invoke({"input": "Hello"})

# With retriever
chain = (
    {"context": retriever | format_docs, "question": RunnablePassthrough()}
    | prompt | model | parser
)