Version: Unreleased 🚧

Custom source

You can extend Botkube functionality by writing custom source plugin. A source allows you to get asynchronous streaming of domain-specific events. For example, streaming Kubernetes or Prometheus events .

Source is a binary that implements the source Protocol Buffers contract.

Goal

This tutorial shows you how to build a custom ticker source that emits an event at a specified interval.

ticker-demo

For a final implementation, see the Botkube template repository.

Prerequisites

Basic understanding of the Go language.
Go at least 1.18.
- See go.mod for the recommended version used by Botkube team.
GoReleaser at least 1.13.

Develop plugin business logic

Create a source plugin directory:

mkdir botkube-plugins && cd botkube-plugins

Initialize the Go module:
```
go mod init botkube-plugins
```

Create the main.go file for the ticker source with the following template:

cat << EOF > main.go
package main

import (
    "context"
    "fmt"
    "time"

    "github.com/hashicorp/go-plugin"
    "github.com/kubeshop/botkube/pkg/api"
    "github.com/kubeshop/botkube/pkg/api/source"
    "gopkg.in/yaml.v3"
)

// Config holds source configuration.
type Config struct {
    Interval time.Duration
}

// Ticker implements the Botkube source plugin interface.
type Ticker struct{}

func main() {
    source.Serve(map[string]plugin.Plugin{
        "ticker": &source.Plugin{
            Source: &Ticker{},
        },
    })
}
EOF

This template code imports required packages and registers Ticker as the gRPC plugin. At this stage, the Ticker service doesn't implement the required Protocol Buffers contract. We will add the required methods in the next steps.

Download imported dependencies:
```
go mod tidy
```
Add the required Metadata method:
```
// Metadata returns details about the Ticker plugin.
func (Ticker) Metadata(_ context.Context) (api.MetadataOutput, error) {
    return api.MetadataOutput{
        Version:     "0.1.0",
        Description: "Emits an event at a specified interval.",
    }, nil
}
```
The Metadata method returns basic information about your plugin. This data is used when the plugin index is generated in an automated way. You will learn more about that in the next steps.
Ąs a part of the Metadata method, you can define the plugin dependencies. To learn more about them, see the Dependencies document.

Add the required Stream method:

// Stream sends an event after configured time duration.
func (Ticker) Stream(ctx context.Context, in source.StreamInput) (source.StreamOutput, error) {
    cfg, err := mergeConfigs(in.Configs)
    if err != nil {
        return source.StreamOutput{}, err
    }

    ticker := time.NewTicker(cfg.Interval)
    out := source.StreamOutput{
        Event: make(chan source.Event),
    }

    go func() {
        for {
            select {
            case <-ctx.Done():
                ticker.Stop()
            case <-ticker.C:
                out.Event <- source.Event{
                    Message: api.NewPlaintextMessage("Ticker Event", true),
                }
            }
        }
    }()

    return out, nil
}

// mergeConfigs merges all input configuration. In our case we don't have complex merge strategy,
// the last one that was specified wins :)
func mergeConfigs(configs []*source.Config) (Config, error) {
    // default config
    finalCfg := Config{
        Interval: time.Minute,
    }

    for _, inputCfg := range configs {
        var cfg Config
        err := yaml.Unmarshal(inputCfg.RawYAML, &cfg)
        if err != nil {
            return Config{}, fmt.Errorf("while unmarshalling YAML config: %w", err)
        }

        if cfg.Interval != 0 {
            finalCfg.Interval = cfg.Interval
        }
    }

    return finalCfg, nil
}

The Stream method is the heart of your source plugin. This method runs your business logic and push events into the out.Output channel. Next, the Botkube core sends the event to a given communication platform.

The Stream method is called only once. Botkube attaches the list of associated configurations. You will learn more about that in the Passing configuration to your plugin section.

Build plugin binaries

Now it's time to build your plugin. For that purpose we will use GoReleaser. It simplifies building Go binaries for different architectures.

note

Instead of GoReleaser, you can use another tool of your choice. The important thing is to produce the binaries for the architecture of the host platform where Botkube is running.

Create the GoReleaser configuration file:

cat << EOF > .goreleaser.yaml
project_name: botkube-plugins
before:
  hooks:
    - go mod download

builds:
  - id: ticker
    binary: source_ticker_{{ .Os }}_{{ .Arch }}

    no_unique_dist_dir: true
    env:
      - CGO_ENABLED=0
    goos:
      - linux
      - darwin
    goarch:
      - amd64
      - arm64
    goarm:
      - 7

snapshot:
  name_template: 'v{{ .Version }}'
EOF

Build the binaries:
```
goreleaser build --rm-dist --snapshot
```

Congrats! You just created your first Botkube source plugin! 🎉

Now it's time to test it locally with Botkube. Once you're done testing, see how to distribute it.

Passing configuration to your plugin

Sometimes your source plugin requires a configuration specified by the end-user. Botkube supports such requirement and provides an option to specify plugin configuration under config. An example Botkube configuration looks like this:

communications:
  "default-group":
    slack:
      channels:
        "default":
          name: "all-teams"
          bindings:
            sources:
              - ticker-team-a
              - ticker-team-b

sources:
  "ticker-team-a":
    botkube/ticker:
      enabled: true
      config:
        interval: 1s
  "ticker-team-b":
    botkube/ticker:
      enabled: true
      config:
        interval: 2m

This means that two different botkube/ticker plugin configurations were bound to the all-teams Slack channel. For each bound configuration Botkube source dispatcher calls Stream method with the configuration specified under the bindings.sources property.

Goal​

Prerequisites​

Develop plugin business logic​

Build plugin binaries​

Passing configuration to your plugin​

Goal

Prerequisites

Develop plugin business logic

Build plugin binaries

Passing configuration to your plugin