ObserverXGetting Started
Installation
Prerequisites
- Node.js version 19 or higher.
- Terminal for accessing the command-line interface (CLI).
ObserverX is considered as a chatbot framework, so it is recommended to have a basic understanding of bot development.
In your project directory, run the following command to install ObserverX:
$ npm install @observerx/core @observerx/console @observerx/database typeorm$ npm install @observerx/core @observerx/console @observerx/database typeorm$ pnpm add @observerx/core @observerx/console @observerx/database typeorm$ pnpm add @observerx/core @observerx/console @observerx/database typeormWhy TypeORM?
TypeORM is an excellent database ORM library for Node.js.
ObserverX uses Postgres as the default database to store chat history and user information with TypeORM. It is currently required to use TypeORM instead of other ORMs.
ObserverX may support other databases / ORMs in the future.
TIP
In the following example, we are going to create a simple command-line interface to chat with the bot. Therefore, we need to install the @observerx/console package.
However, if you want to integrate ObserverX into your project instead, feel free not to install it.
Example: Creating a CLI
ObserverX natively ships a command-line interface (CLI) platform for testing and debugging purposes, and can be installed through @observerx/console.
INFO
A platform is a way to interact with the bot. For example, a CLI platform allows you to chat with the bot in the terminal. There are other platforms such as the HTTP server platform (@observerx/server), the QQ platform (@observerx/qq).
Platforms can be easily built by connecting to target platform's API. ObserverX has rich support for deep customization and extension.
In the project directory you have installed ObserverX, create a file called cli.js with the following content:
import PlatformConsole from '@observerx/console';
import { addEntities, getDataSource } from '@observerx/database';
import ObserverX from '@observerx/core';
addEntities(...ObserverX.getDatabaseEntities());
const dataSource = getDataSource();
await dataSource.initialize();
const platform = new PlatformConsole(dataSource);
platform.start({
model: 'GPT-3.5',
parentId: 'CONSOLE',
prompt: 'default',
});import PlatformConsole from '@observerx/console';
import { addEntities, getDataSource } from '@observerx/database';
import ObserverX from '@observerx/core';
addEntities(...ObserverX.getDatabaseEntities());
const dataSource = getDataSource();
await dataSource.initialize();
const platform = new PlatformConsole(dataSource);
platform.start({
model: 'GPT-3.5',
parentId: 'CONSOLE',
prompt: 'default',
});TIP
Make sure you have "type": "module" set in your package.json to enable the use of await at the top level.
Then, configure your OpenAI credentials in the .env file:
OPENAI_API_KEY=sk-...OPENAI_API_KEY=sk-...Configuring API base URL
By default, ObserverX uses the OpenAI API at https://api.openai.com/v1/. If you want to change it, simply add the following line to your .env file:
OPENAI_API_BASE_URL=<your-api-url>OPENAI_API_BASE_URL=<your-api-url>Finally, run cli.js in your terminal:
$ node cli.js$ node cli.jsIf everything went right, you should be able to chat with GPT-3.5 now.
What's going on?
Let's take a look at the code we just wrote. First, we add the database entities to the database:
import PlatformConsole from '@observerx/console';
import { addEntities, getDataSource } from '@observerx/database';
import ObserverX from '@observerx/core';
addEntities(...ObserverX.getDatabaseEntities());
const dataSource = getDataSource();
await dataSource.initialize();
const platform = new PlatformConsole(dataSource);
platform.start({
model: 'GPT-3.5',
parentId: 'CONSOLE',
prompt: 'default',
});import PlatformConsole from '@observerx/console';
import { addEntities, getDataSource } from '@observerx/database';
import ObserverX from '@observerx/core';
addEntities(...ObserverX.getDatabaseEntities());
const dataSource = getDataSource();
await dataSource.initialize();
const platform = new PlatformConsole(dataSource);
platform.start({
model: 'GPT-3.5',
parentId: 'CONSOLE',
prompt: 'default',
});addEntities is a singleton function that automatically adds the database entities to the database. It should always be called before getDataSource.
getDataSource is a helper function that creates a new DataSource object if none currently exists, or else it returns the currently existing one.
Then, we initialize the data source by dataSource.initialize(). It is required to initialize the data source before calling any ObserverX core-related functions.
import PlatformConsole from '@observerx/console';
import { addEntities, getDataSource } from '@observerx/database';
import ObserverX from '@observerx/core';
addEntities(...ObserverX.getDatabaseEntities());
const dataSource = getDataSource();
await dataSource.initialize();
const platform = new PlatformConsole(dataSource);
platform.start({
model: 'GPT-3.5',
parentId: 'CONSOLE',
prompt: 'default',
});import PlatformConsole from '@observerx/console';
import { addEntities, getDataSource } from '@observerx/database';
import ObserverX from '@observerx/core';
addEntities(...ObserverX.getDatabaseEntities());
const dataSource = getDataSource();
await dataSource.initialize();
const platform = new PlatformConsole(dataSource);
platform.start({
model: 'GPT-3.5',
parentId: 'CONSOLE',
prompt: 'default',
});The code first creates a new PlatformConsole object from the given data source and then starts the console platform with given parameters. Each platform has its own set of parameters, so make sure to check the documentation of the platform you are using.