🎬 langchain-youtube

LangChain retrievers for YouTube — search videos, fetch playlists, extract transcripts, and retrieve video metadata as LangChain Document objects.

Perfect for building RAG applications, video summarizers, and AI-powered YouTube tools.

✨ Features

📦 Installation

pip install langchain-youtube

🔑 Setup (for API key features)

1. Go to the Google Cloud Console
2. Create a project and enable the YouTube Data API v3
3. Create an API key under Credentials

Note: The YouTubeTranscriptRetriever does NOT need an API key!

🚀 Quick Start

Search YouTube Videos

from langchain_youtube import YouTubeSearchRetriever

retriever = YouTubeSearchRetriever(
    api_key="YOUR_API_KEY",
    max_results=5,
    order="relevance",  # or "date", "viewCount", "rating"
)

docs = retriever.invoke("LangChain RAG tutorial")

for doc in docs:
    print(f"📹 {doc.metadata['url']}")
    print(f"   {doc.page_content[:100]}...")
    print(f"   👁 {doc.metadata['view_count']:,} views")

Extract Video Transcripts (No API Key!)

from langchain_youtube import YouTubeTranscriptRetriever

retriever = YouTubeTranscriptRetriever(
    languages=["en", "es"],  # Preferred languages
    chunk_size=3000,          # Split long transcripts for LLM context limits
)

docs = retriever.invoke("https://www.youtube.com/watch?v=VIDEO_ID")

for doc in docs:
    print(f"📝 Chunk {doc.metadata['chunk_index'] + 1}/{doc.metadata['total_chunks']}")
    print(doc.page_content[:200])

Fetch Playlist Videos

from langchain_youtube import YouTubePlaylistRetriever

retriever = YouTubePlaylistRetriever(
    api_key="YOUR_API_KEY",
    playlist_url="https://www.youtube.com/playlist?list=PLxxxxxx",
    max_results=100,
)

docs = retriever.invoke("fetch")  # Query is unused for playlists

for doc in docs:
    print(f"#{doc.metadata['position']} — {doc.page_content[:80]}")

Get Video Metadata

from langchain_youtube import YouTubeVideoRetriever

retriever = YouTubeVideoRetriever(api_key="YOUR_API_KEY")

docs = retriever.invoke("https://www.youtube.com/watch?v=dQw4w9WgXcQ")

video = docs[0]
print(f"Title: {video.page_content}")
print(f"Views: {video.metadata['view_count']:,}")
print(f"Likes: {video.metadata['like_count']:,}")
print(f"Duration: {video.metadata['duration']}")

🔗 Use with LangChain Chains

from langchain_youtube import YouTubeTranscriptRetriever
from langchain_openai import ChatOpenAI
from langchain_core.prompts import ChatPromptTemplate
from langchain_core.output_parsers import StrOutputParser

# Set up retriever and LLM
retriever = YouTubeTranscriptRetriever(languages=["en"])
llm = ChatOpenAI(model="gpt-4o-mini")

prompt = ChatPromptTemplate.from_template(
    "Summarize this YouTube video transcript in 5 bullet points:\n\n{context}"
)

# Build a simple chain
chain = (
    {"context": retriever}
    | prompt
    | llm
    | StrOutputParser()
)

summary = chain.invoke("https://www.youtube.com/watch?v=VIDEO_ID")
print(summary)

🧪 Development

# Clone the repo
git clone https://github.com/urraf/langchain-youtube.git
cd langchain-youtube

# Install in dev mode
pip install -e ".[dev]"

# Run unit tests (no API key needed)
pytest tests/unit_tests/ -v

# Run integration tests (requires API key)
YOUTUBE_API_KEY=your_key pytest tests/integration_tests/ -v

# Lint
ruff check src/ tests/

📄 License

MIT — see LICENSE for details.