feat: add jenkins support

[sinabakh]

The jenkins_diff.sh does relatively the same task as the other diff scripts, here is a list of changes:

• path variable is changed to iac_path
• Remove support for deprecated flags
• Remove post_condition
• Add fail_condition with the percentage_threshold field to fail the pipeline if the change is over the threshold
• Add a function to wrap the diff output in a simple HTML page
• Add few utility functions to make the script work with Jenkins
• Diff-output is always printed in the output
• Save the diff output HTML in a file for further use

Here's a sample Jenkinsfile config:

pipeline {
    agent any
    stages {
        stage('infracost-breakdown') {
            agent {
                docker {
                  image 'infracost-local'
                  args '--user=root --entrypoint='
                }
            }
            environment {
              INFRACOST_API_KEY = credentials('jenkins-infracost-api-key')
              IAC_PATH = 'terraform'
              FAIL_CONDITION = '{"percentage_threshold": 10}'
            }

            steps {
                sh '/scripts/ci/jenkins_diff.sh'

                publishHTML (target: [
                    	allowMissing: false,
      		    	alwaysLinkToLastBuild: false,
      			keepAll: true,
      			reportDir: './',
      			reportFiles: 'infracost_diff_output.html',
      			reportName: "Infracost Diff Output"
    		])

            }
        }
    }
}

Here's the pipeline's "Output consone":

And the output get's its own sub-menu:

[sinabakh]

It would be great if you could take a quick look at it for suggestions or changes before I write the actual documentation.
@alikhajeh1 @aliscott

[alikhajeh1]

Thanks @sinabakh! Will review this today :)

[alikhajeh1]

This is great! To keep the style of fail_condition similar to post_condition, should we also have this option:
'{"never": true}': never fail the Jenkins build. This is the default behavior.

See this comment and the following 2 comments about the pros/cons of why we picked that style.

[sinabakh]

Thanks for the review @alikhajeh1. Regarding the suggestion, I think it can't be applied to such a config. I mean, posting a comment all the time makes sense but failing a build is not something that somebody wants unless a specific condition is met. So adding the never is somehow a duplication.

Let me know if I'm missing a point or it would helo maintain the same standard.

[alikhajeh1]

Good point! Makes sense. What would be the behavior if fail_condition:

1. is not defined? I guess we'd never fail the build so there's no point in having an explicit never option.
2. is set to '{"percentage_threshold": 0}', we'd fail the build if any cost diff is present, so it enables a safety net to be present if the cost spikes a lot or drops a lot (e.g. someone deletes something).

[sinabakh]

That's exactly how you described it.

[alikhajeh1]

@sinabakh thanks for chipping away at this! it's in good shape, just a few minor comments

[alikhajeh1]

I think this can be deleted since env vars are used to run the script (not params passed into the script like GH Action).

[sinabakh]

Sure.

[alikhajeh1]

hmm I wasn't expecting this to default to -1 since for other CI integrations this is the absolute percentage threshold:
"For example, set to 1 to post a comment if the cost estimate changes by more than plus or minus 1%."

Could we do something similar here? I'm actually not sure how useful the drops in cost posting a comment or failing the build are but I lean towards keeping thins consistent until we have more feedback 🤷‍♂️

[sinabakh]

Oh I see, I actually had a different intention. Since setting the default value to 0 will collide with "Fail if there was any change in diff (setting the threshold to 0)", I set its default value as -1. Do you have any suggestions to enhance it?

[alikhajeh1]

So we'd want to handle 2 cases:

1. never fail the build. Default behavior. Maybe we can do that by checking if fail_percentage_threshold is not present?
2. fail the build if the costs increase or decrease by X%, can be done using the fail_percentage_threshold

[sinabakh]

Makes sense, done.

[alikhajeh1]

@aliscott I saw this line in diff.sh and atlantis_diff.sh too, is it redundant since it's not used in this function or am I missing something?

[sinabakh]

Removed it from this script but left the Atlantis for another pr.

[sinabakh]

@alikhajeh1 All done.