youtube_api

The goal of this project is to create an API for fetching the latest videos from YouTube, sorted in reverse chronological order of their publishing date-time, based on a given tag or search query. The response is paginated to efficiently handle large amounts of data.

Running the project

1. Clone the Project: Clone the project repository to your local machine.

git clone https://github.com/lakshya8066/youtube_api.git

2. Navigate to Project Directory: Move to the project directory.

cd youtube_api

3. Start Docker Containers: Run the following command to start MySQL, Elasticsearch, and Kibana containers using Docker Compose.

docker-compose up -d

This command initializes the necessary database and search engine components. 4. Set API Key: Place your YouTube API key in the .env file.

echo "YOUTUBE_API_KEY=<api_key>" > .env

5. Run the Project: Execute the Go application.

go run main.go

API Endpoints

1. Fetch Videos:

• Endpoint: http://localhost:8080/videos?page={page_number}&page_size={page_size}
• Description: Returns a paginated response of videos stored in the MySQL database, sorted in descending order of published datetime.
• Example curl request:

curl --location 'http://localhost:8080/videos?page=2&page_size=10'

2. Update API Key:

• Endpoint: http://localhost:8080/api-key
• Description: Sends a POST request with the new API key when the old key's quota has been exhausted.
• Example curl request:

curl --location --request GET 'http://localhost:8080/api-key' \
--header 'Content-Type: application/json' \
--data '{
    "new_api_key":"anmol"
}'

Architecture

• The application periodically fetches video data from the YouTube Data API v3 every 10 seconds.
• Fetched data is stored in the MySQL server and simultaneously sent to the Elasticsearch server for visualization in the Kibana Dashboard.
• Asynchronous Go routines are employed to handle data retrieval from YouTube, ensuring non-blocking operation.

Optimizations

• Batch Processing: This can be implemented to store multiple videos in a single api call instead of fetching the data multiple times. This will reduce connection calls to the database.

• Removing MySQL: This app could also be build by directly indexing the video data into ElasticSearch as the requirements of the app is minimum.

• Connection Pooling: We can implement connection pooling through which we can reuse the database connections instead of establishing it at each request. This would improve scalability and resource utilization.