@bytebitlabs/nest-openai

nest-openai

OpenAI module for nestjs

Description

OpenAI module for Nest based on the official openai package.

Installation

$ npm i --save @bytebitlabs/nest-openai openai

Usage

Import OpenAiModule:

@Module({
  imports: [OpenAiModule.register({
    apiKey: 'Abc...',
  })],
  providers: [...],
})
export class AiModule {}

Inject OpenAiService:

@Injectable()
export class AiService {
  constructor(private readonly openAiService: OpenAiService) {}
}

Async options

Quite often you might want to asynchronously pass your module options instead of passing them beforehand. In such case, use registerAsync() method, that provides a couple of various ways to deal with async data.

1. Use factory

OpenAiModule.registerAsync({
  useFactory: () => ({
    apiKey: 'Abc...'
  })
});

Obviously, our factory behaves like every other one (might be async and is able to inject dependencies through inject).

OpenAiModule.registerAsync({
  imports: [ConfigModule],
  useFactory: async (configService: ConfigService) => ({
    apiKey: configService.get('OPENAI_API_KEY'),
  }),
  inject: [ConfigService],
}),

2. Use class

OpenAiModule.registerAsync({
  useClass: OpenAiConfigService
});

Above construction will instantiate OpenAiConfigService inside OpenAiModule and will leverage it to create options object.

class OpenAiConfigService implements OpenaiOptionsFactory {
  createOpenAiOptions(): OpenAiModuleOptions {
    return {
      apiKey: 'Abc...'
    };
  }
}

3. Use existing

OpenAiModule.registerAsync({
  imports: [ConfigModule],
  useExisting: ConfigService,
}),

It works the same as useClass with one critical difference - OpenAiModule will lookup imported modules to reuse already created ConfigService, instead of instantiating it on its own.

API Spec

The OpenAiService wraps the Client from the official openai methods. The OpenAiModule.register() takes options object as an argument, read more.