Azure Function ETL using Pandas with CI/CD

This repository hosts a pair of Azure Functions designed for Extract, Transform, Load (ETL) operations:

1. HTTP-triggered Function:
    Handles CSV file uploads from request body to the Blob 'in/input.csv' under container 'hvalfangstcontainer'.

2. Blob-triggered Function:
    Listens for changes to aforementioned Blob 'in/input.csv' and responds by loading the CSV into a Pandas dataframe on changes.
    Conducts calculations to determine correlations between specific columns.
    Stores the resulting correlations in a dictionary and dumps said dictionary to JSON.
    It then proceeds to upload the JSON contents to the Blob 'out/statistics.json'.

A CI/CD pipeline has been implemented utilizing a GitHub Actions Workflow script, which enabled automatic deployment to Azure Function App on repository pushes. Azure resources are provisioned with Terraform via shell scripts 'up' and 'down'.

Requirements

• x86-64
• Linux/Unix
• Python

Creating resources

The shell script 'up' allocates Azure resources with Terraform.

Deleting resources

The shell script 'down' deallocates Azure resources.

Guide

1. Provision Azure Resources

• Run the 'up' script to provision Azure resources with Terraform.

2. Access Azure Portal

• Open your browser and navigate to the Azure Portal.

3. Function App Publish Profile

• Navigate to the newly created Function App 'hvalfangstlinuxfunctionapp'
• Click on 'Get publish profile' to download a file.
• The associated file contents will be used in the next step.

4. GitHub Repository Secrets

• Open the 'Settings' tab of your GitHub repository.
• Click on 'Actions' under 'Security' -> 'Secrets and variables'.
• Create the following repository secret:
  • PUBLISH_PROFILE: Copy value from downloaded file obtained in step #3

5. Deploy Workflow

Our Azure Function will be deployed on repository pushes by utilizing a GitHub Actions workflow script.