The code here is an application with Netlify Functions. These functions are backend and can be uses as APIs. We'll use them as webhooks.
The front end is minimal and the focus is on the functions. It is ready for deployment to Netlify.
The use cases are build around Cloudinary Upload API processing.
The functions in the /functions directory will become your webhooks. You can use them in the scripts found in the /use-cases directory.
The moderation use cases will use:
- webhook_process_google-moderation_queues
- webhook_process_approved_queue
- webhook_process_rejected_queue
The notification use cases uses the SendGrid API to send email. It uses
- webhook_notify_email
Creating optimized video derivatives with transformations involves processing that can take time. It would be helpful to receive an email
with the Cloudinary response showing that the processing completed successfully.
We'll run the video processing as an eager transformation. We'll use the notification_url option to add a webhook. We'll create a Netlify Function, which gives us an API that we can use for a webhook. When the processing is complete, the Netlify Function will call SendGrid's email send function to forward the Cloudinary response to a predefined email address.
If you have deployed the functions/webhook_notify_email function, you can use the URL you have created as the webhook in notification/upload-video-eager script. Execute the upload and check the email specified in your setup for the eager upload response.
We can use the Cloudinary add-on Google Video AI Moderation to examine uploaded video. This add-on looks for content that you might not want to serve to
your website, like pornography or violence. When processing is complete, the add-on will put the video into an approved or rejected queue.
You can view these queues online in the DAM.
This add-on is helpful when we are allowing our users to upload video content. We don't want these videos to be served unless they are marked approved. To
guarantee they can't be served, we upload them with the access_control option containing an access_item with a value of token. As long as
the video has this token, it can't be served.
After the upload is complete, Cloudinary will run it through the Video Moderation process based on options provided with the Upload API command. While this process is running, the video is put in the pending queue. If you look at the Google Moderation Queue at this point, you'll see both of the videos in the pending queue.
We can supply a webhook that will be get called when the moderation step is complete. At that point, the video will be moved to either the approved or rejected queue depending on the analysis. Then the webhook is called.
We'll deploy a Netlify Function to serve as the webhook. This function will call one function to process the approved queue and one to process the rejected queue. If the video is approved, it will be updated using the Admin API update function so that the access_item:token is replaced with access_item:anonymous. This will make it available to be served. If the video is rejected, it will be deleted.
If we look at the Google Moderation Queue at the end of the processing, we'll see that approved video is in the approved queue and we won't find the rejected video as it will have been deleted.
If you look at the moderation queue in the DAM within 60 seconds, you should see both video in the pending queue. The 60 second availability is made possible by a call to sleep in the webhook. This is just for demo purposes.
See NETLIFY_DEPLY.md for instructions on deploying functions to netlify.