Code for CODE4212: Serverless on your own terms with Knative

The talk: https://vmwarecodeconnect.github.io/CodeConnect2020/Evan/

Slides for the talk are in Slides.pdf

Also available on YouTube

The components:

• Image-caption-generator is based on the IBM MAX Image Caption Generator. It is built, unmodified, into a Docker container from the IBM project.

• Captioner is a Spring Boot Java application which reads an image from a POST body, calls the image-caption-generator to determine a caption, and then composites the caption onto the image and returns a new image.

• Image-Extractor is a Node.js application which accepts a com.twitter.tweet CloudEvent from the https://github.com/vaikas/twitter event source, finds the first image in the tweet, fetches it, and returns a new event (com.majordemo.twitter-image) containing the image as binary data.

• configs contains additional Knative object definitions (in particular, trigger definitions) to support the entire workflow.

• The Twitter Source has been extended by adding a sender as well as the existing ContainerSource.

Useful testing commands:

• Captioner: curl -v -H "Content-Type: image/jpg" --data-binary "@./some-image-you-found.jpg" $TARGET -o captioned.jpg

• Image-Extractor: curl -v -H "ce-id: "310437820780216321" -H "ce-source: https://twitter.com/" -H "ce-type: com.twitter.tweet" -H "ce-specversion: 1.0" -H "content-type: application/json" -d "@.\payload.json" $TARGET -o .\temp.jpg

• Tweet-sender: curl -v -H "ce-id: 123" -H "ce-source: curl" -H "ce-type: com.example.test" -H "ce-specversion: 1.0" -H "content-type: image/jpeg" --data-binary "@./some-image-you-found.jpg" $TARGET