@betsys-nestjs/kafka

Kafka library

This library enables messaging using Apache Kafka.

Dependencies

Package	Version
kafkajs	^2.0.0
@kafkajs/confluent-schema-registry	^3.0.0
reflect-metadata	^0.1.12
rxjs	^7.1.0

Usage

• To start using this library simply import KafkaModule to your module.

@Module({
    imports: [
        KafkaModule.forFeature(kafkaConfig(), 'my-handle'),
    ]
})
export class AppModule {
    // ...
}

• Pass these arguments to forFeature:
• kafkaConfig:

const kafkaConfig: KafkaModuleConfig = {
    brokers: ['localhost:29092'], // list of kafka broker hosts
    clientId: 'test-client-id', // unique client identification
    registryHost: 'http://localhost:8081', // schema registry host
};

• dbHandle - unique handle identifier

• NOTE: Library requires @betsys-nestjs/logger to work. It's planned to decouple this library of this dependency.

• setup infrastructure

    await schemaRegistryProvider.createSchemaRegistry();
    await kafkaConnectionUtils.createKafkaConnection();
    await kafkaConnectionUtils.connectProducer();

• create schema e.g. dog.schema.avro

{
    "name": "Dog",
    "type": "record",
    "namespace": "test_namespace",
    "fields": [
        {
            "name": "name",
            "type": "string"
        },
        {
            "name": "age",
            "type": "int"
        },
        {
          "name": "birthplace",
          "type": [
            "null",
            "string"
          ],
          "default": null
        }
    ]
}

• register schema to retrieve its schema ID
  • subject name must have -value suffix

    const dogSchema = await readAVSCAsync(path.join(__dirname, 'dog.schema.avsc'));
    const schemaId = await schemaRegistryProvider.registerSchema(dogSchema, {
        subject: `dog-subject-value`,
        compatibility: COMPATIBILITY.BACKWARD,
    });

• create topic

    await kafkaConnectionUtils.createTopic('dog-topic');

• encode message using schema ID

    const payload = await schemaRegistryProvider
        .getSchemaRegistry()
        .encode(schemaId, { 
            name: 'Buddy',
            age: 6,
            birthplace: 'Děčín',
        });

• produce message

    await kafkaConnectionUtils.send('dog-topic', [{ value: payload }])

---

OPTIONAL

• create consumer and callback to represent logic how to process messages

    await kafkaConnectionUtils.connectConsumer({ groupId: 'unique-consumer-group' })

    const createCallback = async (
        schemaProvider: SchemaRegistryProvider,
        schema: RawAvroSchema,
    ): Promise<(payload: EachMessagePayload) => Promise<void>> => async (payload: EachMessagePayload): Promise<void> => {
        const schemaRegistry = schemaRegistryProvider.getSchemaRegistry();
        const decodedMessage = await schemaRegistry.decode(
            payload.message.value as Buffer,
            {
                [SchemaType.AVRO]: { readerSchema: schema },
            },
        );
    
        console.log(decodedMessage);
    };

• subscribe to topic and run consumer

    await kafkaConnectionUtils.subscribeAndRunConsumer(
        { topics: ['dog-topic'] },
        await createCallback(
            schemaRegistryProvider,
            dogSchema,
        ),
    );

Logger

The library is ready to work with logger. To enable it you need to implement your own logger service based on abstraction provided by this library.

You can simply implement custom service following KafkaLoggerInterface.

Example using @betsys-nestjs/logger:

import { Injectable } from '@nestjs/common';
import { Logger as NestLogger } from '@betsys-nestjs/logger';
import { Logger } from '@betsys-nestjs/postgres';

@Injectable()
export class KafkaLogger implements KafkaLoggerInterface {
    constructor(private readonly logger: NestLogger) {}

    info(message: string): void {
        // eslint-disable-next-line no-console
        this.logger.info(message);
    }

    setContext(context: string): void {
        this.logger.setContext(context);
    }
}

In setContext you can define some context for further logging. info method is responsible for logging itself so you can either use some console.log or any logger based on your preference like winston etc.

To start using Logger service, you simply insert class references to forFeature method of KafkaModule like this:

KafkaModule.forFeature({
    ...kafkaConfig(), 
    logger: KafkaTestLogger,
})