jmcdo29
@ogma/nestjs-module
A NestJS module for the Ogma logging package.
Installation
Installation is pretty simple, just npm i @ogma/nestjs-module or yarn add @ogma/nestjs-module
Usage
The OgmaService is a SINGLETON scoped service class in NestJS. That being said, if you want a new instance for each service, you can use the OgmaModule.forFeature() method and the @OgmaLogger() decorator. When working with OgmaModule.forFeature() you can pass an object as the second parameter to determine if you want the logger to be request scoped or not. This object looks like { addRequestId: true|false }.
Ogma is a lightweight logger with customization options, that prints your logs in a pretty manner with timestamping, colors, and different levels. See the GitHub repository for Ogma to learn more about configuration and options.
Configuration
In your root module, import OgmaModule.forRoot or OgmaModule.forRootAsync to apply application wide settings, such as color, json, and application. You can also set a default context as a fallback if no context is passed in the forFeature static method. You can use the standard method of async module configuration meaning useClass, useValue, and useFactory are all available methods to work with.
Synchronous configuration
import { APP_INTERCEPTOR } from '@nestjs/core';
import { OgmaInterceptor, OgmaModule } from '@ogma/nestjs-module';
import { ExpressParser } from '@ogma/platform-express';
@Module({
imports: [
OgmaModule.forRoot({
service: {
color: true,
json: false,
application: 'NestJS'
},
interceptor: {
http: ExpressParser,
ws: false,
gql: false,
rpc: false
}
})
],
providers: [
{
provide: APP_INTERCEPTOR,
useClass: OgmaInterceptor
}
]
})
export class AppModule {}Asynchronous configuration
import { ConfigService } from '@nestjs/config';
import { APP_INTERCEPTOR } from '@nestjs/core';
import { OgmaInterceptor, OgmaModule } from '@ogma/nestjs-module';
import { ExpressParser } from '@ogma/platform-express';
import { appendFile } from 'fs';
@Module({
imports: [
OgmaModule.forRootAsync({
useFactory: (config: ConfigService) => ({
service: {
json: config.isProd(),
stream: {
write: (message) => {
appendFile(config.getLogFile(), message, (err) => {
if (err) {
throw err;
}
return true;
});
}
},
application: config.getAppName()
},
interceptor: {
http: ExpressParser,
ws: false,
gql: false,
rpc: false
}
}),
inject: [ConfigService]
})
],
providers: [
{
provide: APP_INTERCEPTOR,
useClass: OgmaInterceptor
}
]
})
export class AppModule {}From here in each module you need to add the OgmaService you can add OgmaModule.forFeature(context), where context is the context you want to add to the log. This context can be either a class object (not an instance) or a string.
@Module({
imports: [OgmaModule.forFeature(MyService)], // or OgmaModule.forFeature(MyService.name)
providers: [MyService]
})
export class MyModule {}@Injectable()
export class MyService {
constructor(
@OgmaLogger(MyService) private readonly logger: OgmaService // or @OgmaLogger(MyService.name)
) {}
// ...
}The above would set up the OgmaService to add the context of [MyService] to every log from within the MyModule module.
Request Scoping the Logger
As mentioned above, you can pass a second parameter to the forFeature method to tell Ogma that you want this logger to be request scoped and add a request Id to the logs. This request Id is generated in the interceptor which is important to note in the case of a request that fails at the Guard or Middleware level, as they will not yet have this ID. You can choose to add your own middleware to create an id if you so choose and retrieve it later. There's also a new decorator for request scoped loggers @OgmaLoggerRequestScoped(). This decorator acts exactly like the @OgmaLogger() decorator, with the same parameters and all, it just uses a different injection token with the form of OGMA_REQUEST_SCOPED_SERVICE:<Service_Name>.
Warning: Please make sure you understand the implications of using a request scoped service!
@Module({
imports: [OgmaModule.forFeature(MyService, { addRequestId: true })],
providers: [MyService]
})
export class MyModule {}@Injectable()
export class MyService {
constructor(
@OgmaLoggerRequestScoped(MyService)
private readonly logger: OgmaService
) {}
// ...
}forFeature/forFeatures
Normally to create a new provider with a separate context you'll use theforFeature method and pass in a string or a class. e.g. OgmaModule.forFeature(FooService). This will create the proper injection token for @OgmaLogger(FooService).
There is also the forFeatures method which takes in a parameter for multiple features to be created. This parameter can be a mixed array of classes, strings, and objects, where the objects have a shape of { contexts: string | (() => any) | Type<any>, options: OgmaProviderOptions }. This is useful for when you want to register multiple OgmaServices in the same module, such as one logger for the Service and one for the Controller, or for a Service and a Filter.
OgmaInterceptor
Ogma also comes with a built in Interceptor for logging requests made to your server. The interceptor is by default not bound to the server. To bind the interceptor, you can either use @useInterceptor on the route handlers you want to automatically log, or use the global binding with the provider {provide: APP_INTERCEPTOR, useClass: OgmaInterceptor}. You can also use app.useGlobalInterceptors(), but this is not encouraged, as the interceptor requires several dependencies.
The interceptor will need to be told what parsers it should be using for each type of request that can be intercepted. By default, all of these values are set to false, and if you try to bind the interceptor without setting a parser, you will end up with errors when running the server. If you'd like to know more about why this is the default behavior, please look at the interceptor design decisions part of the docs. Below is the general form that the interceptor logs will take:
[ISOString TimeStamp] [Application Name] PID RequestID [Context] [LogLevel]| Remote-Address - method URL protocol Status Response-Time ms - Response-Content-LengthThis request ID is generated inside the OgmaInterceptor currently by using Math.random(). You may extend the logger and modify the generateRequestId method to change to it be however you like though.
Where Context is the class-method combination of the path that was called. This is especially useful for GraphQL logging where all URLs log from the /graphql route.
If you would like to skip any request url path, you can pass in a decorator either an entire class or just a route handler with the @OgmaSkip() decorator.
Note: Be aware that as this is an interceptor, any errors that happen in middleware, such as Passport's serialization/deserialization and authentication methods through the PassportStrategy will not be logged in the library. You can use an ExceptionFilter to manage that. The same goes for guards due to the request lifecycle
OgmaInterceptor Configuration Options
All configuration options are just that: options. None of them need to be provided, but it could prove useful to do in certain cases. Find the table of values below to have a better idea of what is necessary to provide the options.
| name | value | default | description |
|---|---|---|---|
| gql | false or AbstractInterceptorService | false | The GraphQL parser for the OgmaInterceptor |
| http | false or AbstractInterceptorService | false | The HTTP parser for the OgmaInterceptor |
| rpc | false or AbstractInterceptorService | false | The Microservice parser for the OgmaInterceptor |
| ws | false or AbstractInterceptorService | false | The Websocket parser for the OgmaInterceptor |
Each of the above options, as mentioned, is false meaning that the requests for that type would not be parsed by the OgmaInterceptor and the request would flow as normal.
Note: The
AbstractInterceptorServiceis the class that allOgmaInterceptorServiceparsers should extend, in order to adhere to the call scheme of the built inDelegatorService. This allows for easy extension and overwriting of the parsers.
Interceptor Design Decisions
Due to the incredible complex nature of Nest and its DI system, there needed to be some sort of way to tell users at bootstrap that if the interceptor is to be used, which should be the default behavior, then it should have one of the @ogma/platform-* packages installed, or a custom parser should be provided. Every custom parser should extend the AbstractInterceptorService to ensure that A) Typescript doesn't complain about mismatched types, and B) the DelegatorService which handles the calls to each parser, can be sure it is getting back what it expects. If you are really, really sure about what you are doing, you can always override the setting with as any to remove the Typescript warnings, but use that at your own risk.
The interceptor was designed to be adaptable, and to be able to work with any context thrown at it, but only if the parser for that context type is installed. The most common parser would be @ogma/platform-express, which will work for HTTP requests with the Express server running under the hood (Nest's default). All other parsers provided by the @ogma namespace follow a similar naming scheme, and are provided for what Nest can use out of the box (including microservices named in the microservices chapter of the Nest docs.)
Now, for the reasoning that all parsers are defaulted to false is to A) ensure that the developer does not expect the interceptor to work out of the box, B) ensure that the developer is aware of what parser is being used, and C) ensure that the parser(s) being used are installed without being blindly used (this means Typescript will complain if the class doesn't exist, whereas with JavaScript it may be okay if a linter is not installed).
Extending Pre-Built Parsers
As the pre-built parsers are built around Object Oriented Typescript, if you want to change the functionality of one of the pre-built parsers, you can always create a new class that extends the class, change the specific method(s) you want, and then provide that as your parser. All the other methods should still come from the base class and not be affected.
Putting it All Together
Okay, so now we're ready to add the OgmaModule to our Application. Let's assume we have a CatsService, CatsController and CatsModule and a ConfigService and ConfigModule in our Application. Let's also assume we want to use a class to asynchronously configure out OgmaModule. For now, assume the methods exist on the ConfigService. Let's also assume we want to log things in color to our process.stdout.
import { FastifyParser } from '@ogma/platform-fastify';
import { OgmaModuleOptions } from '@ogma/nestjs-module';
@Injectable()
export class OgmaModuleConfig {
constructor(private readonly configService: ConfigService) {}
createModuleConfig(): OgmaModuleOptions {
return {
service: {
// returns one of Ogma's log levels, or 'ALL'.
logLevel: this.configService.getLogLevel(),
color: true,
// could be something like 'MyAwesomeNestApp'
application: this.configService.getAppName()
},
interceptor: {
http: FastifyParser
}
};
}
}Next, in our AppModule we can import the OgmaModule like so
@Module({
imports: [
CatsModule,
ConfigModule,
OgmaModule.forRootAsync({
// configuration class we created above
useClass: OgmaModuleConfig,
imports: [ConfigModule]
})
]
})
export class AppModule {}And now we have the interceptor bound and an OgmaService instance created for the application. If we want to add the OgmaService as the general logger for Nest we can do the following in our main.ts
async function bootstrap() {
const app = await NestFactory.create(AppModule, { logger: false });
const logger = app.get<OgmaService>(OgmaService);
app.useLogger(logger);
await app.listen(3000);
}
bootstrap();With this, you will lose the initial logs about instantiation of the modules, but you will still get messages about what routes are bootstrapped and when the server is ready.
If we wanted to add the logger to the CatsService we can use the module's forFeature static method to give the logger service its own context variable.
@Module({
imports: [OgmaModule.forFeature(CatsService)],
controllers: [CatsController],
service: [CatsService]
})
export class CatsModule {}And now in the CatsService the OgmaService can be injected like so:
@Injectable()
export class CatsService {
constructor(
@OgmaLogger(CatService) private readonly logger: OgmaService
) {}
}And now this.logger is available in your CatsService class.
Demo
Below is what the OgmaInterceptor can do. These are the logs I usually see during the integration testing, and show off just what is capable in terms of the metadata captured on requests.
Get in Touch
If there is something that needs to be addressed in regards to the module, feel free to make an issue. If you are looking to contact me, you can either email me, or find me on discord as PerfectOrphan31#6003.
Contributing
Please make sure that any and all pull requests follow the code standards enforced by the linter and prettier. Otherwise, any and all pull requests are welcomed, though an issue created to first track the PR would be appreciated. Feel free to make suggestions on how the module can improve and what can be done better.
License
NestJS-Ogma has an MIT License.