GitHub Action Job Runtime Data Extractor

extractor.rb

require 'date'
require 'json'

def get_seconds_difference(start, finish)
    ((finish - start) * 24 * 60 * 60).to_i
end

# from gh run list --branch main --event push --status completed --json displayTitle,headSha,name,databaseId,startedAt,updatedAt,workflowName --limit 1000 > run_list.json
data = JSON.parse(File.read('./run_list.json'))

puts "Workflow Name,Job Name,Start Time,Finish Time,Elapsed Time"
for run in data do
    run_details = JSON.parse(`gh run view #{run['databaseId']} --json name,workflowName,jobs,createdAt,updatedAt`)
    for job in run_details['jobs'] do
        start = DateTime.parse(job['startedAt'])
        finish = DateTime.parse(job['completedAt'])
        puts "\"#{run_details['workflowName']}\",\"#{job['name']}\",#{start},#{finish},#{get_seconds_difference(start, finish)}"
    end
end

github_action_job_runtime_data_extractor.md

This is a quick and dirty script to get some data about GitHub action job running time for ad hoc analysis. It uses the GitHub CLI rather than the real API library because it simplified auth and made experimentation super simple and easy.

I found the JSON data didn't give accurate results, though it's not clear how or why. The "unstructured" version of the code uses the non-JSON data and parses the strings to get the time data.

unstructured_extractor.rb

require 'date'
require 'json'

def get_seconds(minutes, seconds)
    mins = if minutes.nil?
        0
    else
        minutes.chop().to_i
    end

    mins * 60 + seconds.chop().to_i
end

# from gh run list --branch main --event push --status completed --json displayTitle,headSha,name,databaseId,startedAt,updatedAt,workflowName --limit 500 > run_list.json
data = JSON.parse(File.read('./run_list.json'))

puts "Workflow Name,Job Name,Duration"
for run in data do
    run_details = `gh run view #{run['databaseId']}`.split("\n").map(&:strip)
    for line in run_details do
        if /^[^ ]+ (?<job>.*) in (?<minutes>\d+m)?(?<seconds>\d{1,2}s) \(ID \d+\)$/ =~ line
            puts "\"#{run['workflowName']}\",\"#{job}\",#{get_seconds(minutes, seconds)}"
        end
    end
end