gitlab_artifact_remover

#!/usr/bin/env ruby

# Inspired by the bash script by Chris Arceneaux
# https://gist.github.com/carceneaux/b75d483e3e0cb798ae60c424300d5a0b

require 'gitlab'
require 'thor'
require 'time'
require 'uri'

class GitLabArtifactRemover < Thor
  class_option :endpoint, type: :string, required: true, aliases: '-e', desc: 'GitLab API endpoint (e.g., https://gitlab.example.com/api/v4)'
  class_option :token, type: :string, required: true, aliases: '-t', desc: 'GitLab access token'
  class_option :project, type: :string, required: true, aliases: '-p', desc: 'Project ID or path (e.g., my-group/my-project)'

  desc 'remove', 'Remove job artifacts from a GitLab project'
  option :older_than, type: :string, aliases: ['-o', '--older-then'], desc: 'Only remove artifacts older than this duration (e.g., 3d, 1w, 24h). Without this flag, all artifacts are targeted.'
  option :force, type: :boolean, default: false, aliases: '-f', desc: 'Skip confirmation prompt'
  def remove
    configure_client

    cutoff_time = nil
    if options[:older_than]
      cutoff_time = parse_duration(options[:older_than])
      unless cutoff_time
        puts "Error: Invalid --older-than format. Use formats like 3d, 1w, 24h, 30m."
        exit 1
      end
      puts "Removing artifacts older than #{options[:older_than]} (before #{cutoff_time.iso8601})"
    end

    puts "Fetching jobs from project: #{options[:project]}"

    all_jobs = []
    @client.jobs(options[:project], per_page: 100).auto_paginate do |job|
      all_jobs << job
    end

    puts "Total jobs found: #{all_jobs.count}"

    jobs_with_artifacts = all_jobs.select do |job|
      has_real_artifacts = job.artifacts.any? { |artifact| artifact.file_type == 'archive' }
      next false unless has_real_artifacts

      if cutoff_time
        job_time = parse_time(job.finished_at) || parse_time(job.created_at)
        next false unless job_time
        job_time < cutoff_time
      else
        true
      end
    end

    if jobs_with_artifacts.empty?
      puts "No jobs with artifacts found matching criteria."
      return
    end

    puts "\n#{jobs_with_artifacts.count} job(s) with artifacts found:"
    jobs_with_artifacts.each do |job|
      time_str = job.finished_at || job.created_at
      puts "  - Job #{job.id} (#{job.name}, ref: #{job.ref}) – #{time_str}"
    end

    unless options[:force]
      print "\nDo you want to delete artifacts for these jobs? (yes/no): "
      response = STDIN.gets.chomp.downcase

      unless %w[yes y].include?(response)
        puts "Operation cancelled."
        exit 0
      end
    end

    puts "\nDeleting artifacts..."
    deleted_count = 0
    failed_count = 0

    encoded_project = URI.encode_www_form_component(options[:project].to_s)

    jobs_with_artifacts.each do |job|
      begin
        @client.delete("/projects/#{encoded_project}/jobs/#{job.id}/artifacts")
        puts "  ✓ Deleted artifacts for Job #{job.id} (#{job.name})"
        deleted_count += 1
      rescue Gitlab::Error::Error => e
        puts "  ✗ Failed to delete artifacts for Job #{job.id} (#{job.name}): #{e.message}"
        failed_count += 1
      end
    end

    puts "\nDone. Deleted: #{deleted_count}, Failed: #{failed_count}"

  rescue Gitlab::Error::Error => e
    puts "Error: #{e.message}"
    exit 1
  end

  private

  def configure_client
    @client = Gitlab.client(
      endpoint: options[:endpoint],
      private_token: options[:token]
    )
  rescue => e
    puts "Error configuring GitLab client: #{e.message}"
    exit 1
  end

  def parse_duration(str)
    return nil unless str =~ /^(\d+)([dhwm])$/i

    num = $1.to_i
    unit = $2.downcase
    now = Time.now

    case unit
    when 'd' then now - (num * 24 * 60 * 60)
    when 'h' then now - (num * 60 * 60)
    when 'w' then now - (num * 7 * 24 * 60 * 60)
    when 'm' then now - (num * 60)
    else nil
    end
  end

  def parse_time(str)
    return nil if str.nil? || str.to_s.empty?

    Time.parse(str.to_s)
  rescue ArgumentError
    nil
  end
end

GitLabArtifactRemover.start(ARGV)