InvoiceMe

---

A cloud-native app to automatically generate invoices from the Harvest timesheet application and notify via Slack.

Why

Creating invoices can be tricky. It requires:

• Remembering to create and send the invoice in the first place
• Manually looking at the number of days worked on a timesheet application (Harvest)
• Ensure the information is accurate (prone to human error)
• etc.

Enter InvoiceMe

InvoiceMe is a python lambda function that combines static invoice data (a mix of config files and state in a dynamodb table) with the dynamic timesheet content from Harvest to produce an invoice. The lambda builds a json payload for the invoice and makes use of invoiced to generate the PDF. This is then uploaded to S3 and a slack notification with a signed URL to the file is sent to the user. The app uses the following AWS services:

• Lambda - for running serverless functions
• DynamoDB - for keeping a tiny bit of state used to generate invoices
• Cloudwatch - for scheduling the lambda to run automatically
• IAM - for permissions
• S3 - for long term invoice storage
• Secrets Manager - for storing api keys

Deploying

I've supplied a bunch of terraform templates to simplify the deployment of the app. All you have to do is:

1. Edit secretsmanager.tf and enter your Harvest API token & Slack webhook
2. Edit lambda.tf and enter your Harvest account ID in the environment variables
3. Run the following commands:

./terraform init
./terraform apply

(note: the init is important as I use the archive provider to zip the lambda src on the fly, so it'll fail if you don't have the provider setup).

Once deployed, when the lambda runs you'll recieve a slack notification like so: ...which links you to your shiny new professional-looking PDF:

Considerations

• The code could probably do with being split up and just made a bit nicer in general. The scope of the app changed as I was going along so the code could be cleaned up. Error handling is lacking...