lance / twitter-translate-orchestration

5 minute lightning talk on building a Twitter translation bot with Knative Functions

KubeCon NA 2022 - Build a Twitter Translate Bot in 5 Minutes

This demo converts tweets into English using Camel-K and Knative.

It uses a Camel-K Kamelet to periodically search Twitter for [some value]. The Kamelet converts each tweet into a CloudEvent, sending it to the default Knative Broker. A translate function receives these events via a Trigger between the Broker and the function which filters for these types of events.

The translate function inspects the tweet language and if it's English, it simply returns the tweet text as the data for a translation event. If it's not English, the Google Translation API is used to convert the tweet text to English, and the result is returned as the data in the translation event.

A viewer function is deployed to listen for translation events and display them in the logs.

Prerequisites

• A cluster with Knative installed
• CamelK https://camel.apache.org/camel-k/1.9.x/installation/installation.html installed
• Google Translation API key in translate/service-account-file.json
• Update the Twitter API credentials in ./resources/twitter-search-source-binding.yaml

Steps

• Deploy the translate and viewer functions

func deploy -p viewer
func deploy -p translate

• Install a default Knative Broker

kn broker create default

• Create the Tweet->Translate trigger

kn trigger create twitter-trigger -s translate --filter type=twitter.search.source

• Create the Translation->Viewer trigger

kn trigger create translate-trigger -s viewer --filter type=knative.function.translation

• Install the Kamelet

kubectl apply -f resources/twitter-search-source-binding-v1.yaml

• Check the logs

k logs -l app=viewer --tail 10 -f

• Cleanup

./resources/teardown.sh