YugaPlus: Movie Recommendations Service with OpenAI, Spring AI, and PostgreSQL pgvector

This is a sample movie recommendation service that is built on an OpenAI embedding model, Spring AI framework and PostgreSQL pgvector.

The service takes user questions written in plain English and uses a gen AI stack (OpenAI, Spring AI and PostgreSQL pgvector) to provide the user with the most relevant movie recommendations.

Prerequisites

The latest version of Docker and Docker Compose.
OpenAI API key

If you're planning to run the app on bare metal, then make sure to have:

Node.js 20+
Java 21+. Use sdkman to install it within a minute.
Maven 3.9+

Start Database Instance In Docker

The pgvector extension is supported by a single-server PostgreSQL instance as well as a multi-node YugabyteDB cluster. Feel free to use any of the database options.

Start PostgreSQL in Docker

Create the postgres-volume directory for the Postgres container's volume in your home dir. The volume is handy if you'd like to access the logs easily and don't want to lose data when the container is recreated from scratch:
```
mkdir ~/postgres-volume
```
Create a custom Docker network:
```
docker network create yugaplus-network
```

Start the Postgres container with the pgvector extension:

docker run --name postgres --net yugaplus-network \
    -e POSTGRES_USER=postgres -e POSTGRES_PASSWORD=password \
    -p 5432:5432 \
    -v ~/postgres-volume/:/var/lib/postgresql/data \
    -d ankane/pgvector:latest

Make sure the container is running:
```
docker container ls -f name=postgres
```

Start YugabyteDB in Docker

Create a custom Docker network:
```
docker network create yugaplus-network
```

Start a YugabyteDB cluster:

rm -r ~/yugabyte-volume
mkdir ~/yugabyte-volume

docker run -d --name yugabytedb-node1 --net yugaplus-network \
-p 15433:15433 -p 7001:7000 -p 9001:9000 -p 5433:5433 \
-v ~/yugabyte-volume/node1:/home/yugabyte/yb_data --restart unless-stopped \
yugabytedb/yugabyte:latest \
bin/yugabyted start --base_dir=/home/yugabyte/yb_data --daemon=false

# Wait until the first node is initialized and ready to accept connection
while ! docker exec -it yugabytedb-node1 postgres/bin/pg_isready -U yugabyte -h yugabytedb-node1; do sleep 1; done

docker run -d --name yugabytedb-node2 --net yugaplus-network \
-p 15434:15433 -p 7002:7000 -p 9002:9000 -p 5434:5433 \
-v ~/yugabyte-volume/node2:/home/yugabyte/yb_data --restart unless-stopped \
yugabytedb/yugabyte:latest \
bin/yugabyted start --join=yugabytedb-node1 --base_dir=/home/yugabyte/yb_data --daemon=false
    
docker run -d --name yugabytedb-node3 --net yugaplus-network \
-p 15435:15433 -p 7003:7000 -p 9003:9000 -p 5435:5433 \
-v ~/yugabyte-volume/node3:/home/yugabyte/yb_data --restart unless-stopped \
yugabytedb/yugabyte:latest \
bin/yugabyted start --join=yugabytedb-node1 --base_dir=/home/yugabyte/yb_data --daemon=false

Start Application

You have an option of deploying the application in Docker or on your host operating system (bare metal).

Start Application With Docker Compose

By default, the containers will attempt to connect to the PostgreSQL container. If you use YugabyteDB:

Update the following settings in the docker-compose.yaml:

- DB_URL=jdbc:postgresql://yugabytedb-node1:5433/yugabyte
- DB_USER=yugabyte
- DB_PASSWORD=yugabyte

If you use YugabyteDB 2.20.1 or later, then add the following parameter:

- DB_CONN_INIT_SQL="SET yb_silence_advisory_locks_not_supported_error=true"

During the first run, build an image and only then start the containers:

docker-compose up --build

Once the image is ready, use this command to start the containers:

docker-compose up

Start Application in Docker

Start the backend in Docker:

Create a Docker image for the backend:

cd backend
docker build -t yugaplus-backend .

Start a backend container:

For PostgreSQL:

docker run --name yugaplus-backend --net yugaplus-network -p 8080:8080 \
    -e DB_URL=jdbc:postgresql://postgres:5432/postgres \
    -e DB_USER=postgres \
    -e DB_PASSWORD=password \
    -e OPENAI_API_KEY=your-api-key \
    yugaplus-backend

For YugabyteDB:

docker run --name yugaplus-backend --net yugaplus-network -p 8080:8080 \
    -e DB_URL=jdbc:postgresql://yugabytedb-node1:5433/yugabyte \
    -e DB_USER=yugabyte \
    -e DB_PASSWORD=yugabyte \
    -e OPENAI_API_KEY=your-api-key \
    yugaplus-backend

Start the frontend in Docker:

Create a Docker image for the frontend:

cd frontend
docker build -t yugaplus-frontend .

Start a frontend container:

docker run --name yugaplus-frontend --net yugaplus-network -p 3000:3000 \
    -e REACT_APP_PROXY_URL=http://yugaplus-backend:8080 \
    yugaplus-frontend

Start Application on Bare Metal

By default, the backend will attempt to connect to the PostgreSQL container. If you use YugabyteDB:

Update the following settings in the application.properties file:

spring.datasource.url = jdbc:postgresql://yugabytedb-node1:5433/yugabyte
spring.datasource.username = yugabyte
spring.datasource.password = yugabyte

If you use YugabyteDB 2.20.1 or later, then add the following parameter:

spring.datasource.hikari.connection-init-sql="SET yb_silence_advisory_locks_not_supported_error=true"

Start the backend:

Provide your OpenAI key in the backend/src/main/resources/application.properties file:
```
spring.ai.openai.api-key=sk-
```
Go to the backend directory and start the app:
```
cd backend
mvn spring-boot:run
```

Start the frontend:

Go to the frontend directory:
```
cd frontend
```
Start the app:
```
npm install
npm start
```

Test Application

Go to http://localhost:3000 and test the application.

user1@gmail.com/password - already has some movies in the library
user2@gmail.com/password - the library is empty

Try a few prompts: A movie about a space adventure. A kids-friendly movie with unexpected ending.

AngelCruzL/YugaPlus