gitlab-ce/scripts/static-analysis

#!/usr/bin/env ruby
# frozen_string_literal: true

# We don't have auto-loading here
require_relative '../lib/gitlab'
require_relative '../lib/gitlab/popen'
require_relative '../lib/gitlab/popen/runner'

class StaticAnalysis
  # `ALLOWED_WARNINGS` moved to scripts/allowed_warnings.txt

  Task = Struct.new(:command, :duration) do
    def cmd
      command.join(' ')
    end
  end
  NodeAssignment = Struct.new(:index, :tasks) do
    def total_duration
      return 0 if tasks.empty?

      tasks.sum(&:duration)
    end
  end

  def self.project_path
    project_root = File.expand_path('..', __dir__)

    if Gitlab.jh?
      "#{project_root}/jh"
    else
      project_root
    end
  end

  # `gettext:updated_check` and `gitlab:sidekiq:sidekiq_queues_yml:check` will fail on FOSS installations
  # (e.g. gitlab-org/gitlab-foss) since they test against a single
  # file that is generated by an EE installation, which can
  # contain values that a FOSS installation won't find. To work
  # around this we will only enable this task on EE installations.
  TASKS_WITH_DURATIONS_SECONDS = [
    Task.new(%w[yarn run lint:prettier], 270),
    Task.new(%w[bin/rake gettext:lint], 120),
    Task.new(%W[scripts/license-check.sh #{project_path}], 80),
    (Gitlab.ee? ? Task.new(%w[bin/rake gettext:updated_check], 60) : nil),
    Task.new(%w[bin/rake lint:static_verification], 45),
    Task.new(%w[bin/rake gitlab:sidekiq:all_queues_yml:check], 30),
    (Gitlab.ee? ? Task.new(%w[bin/rake gitlab:sidekiq:sidekiq_queues_yml:check], 25) : nil),
    Task.new(%w[bin/rake config_lint], 10),
    Task.new(%w[yarn run internal:stylelint], 10),
    Task.new(%w[scripts/lint-conflicts.sh], 10),
    Task.new(%w[scripts/lint-vendored-gems.sh], 10),
    Task.new(%w[yarn run check-dependencies], 1),
    Task.new(%w[scripts/gemfile_lock_changed.sh], 1),
    Task.new(%w[yarn run deps:check:all --no-cache], 60)
  ].compact.freeze

  def run_tasks!(options = {})
    total_nodes = (ENV['CI_NODE_TOTAL'] || 1).to_i
    current_node_number = (ENV['CI_NODE_INDEX'] || 1).to_i
    node_assignment = tasks_to_run(total_nodes)[current_node_number - 1]

    if options[:dry_run]
      puts "Dry-run mode!"
      return
    end

    static_analysis = Gitlab::Popen::Runner.new
    start_time = Time.now
    static_analysis.run(node_assignment.tasks.map(&:command)) do |command, &run|
      task = node_assignment.tasks.find { |task| task.command == command }
      puts
      puts "$ #{task.cmd}"

      result = run.call

      puts "==> Finished in #{result.duration} seconds (expected #{task.duration} seconds)"
      puts
    end

    puts
    puts '==================================================='
    puts "Node finished running all tasks in #{Time.now - start_time} seconds (expected #{node_assignment.total_duration})"
    puts
    puts

    if static_analysis.all_success_and_clean?
      puts 'All static analyses passed successfully.'
    elsif static_analysis.all_success?
      puts 'All static analyses passed successfully with warnings.'
      puts

      emit_warnings(static_analysis)

      # We used to exit 2 on warnings but `fail_on_warnings` takes care of it now.
    else
      puts 'Some static analyses failed:'

      emit_warnings(static_analysis)
      emit_errors(static_analysis)

      exit 1
    end
  end

  def emit_warnings(static_analysis)
    static_analysis.warned_results.each do |result|
      warn
      warn "**** #{result.cmd.join(' ')} had the following warning(s):"
      warn
      warn result.stderr
      warn
    end
  end

  def emit_errors(static_analysis)
    static_analysis.failed_results.each do |result|
      puts
      puts "**** #{result.cmd.join(' ')} failed with the following error(s):"
      puts
      puts result.stdout
      puts result.stderr
      puts
    end
  end

  def tasks_to_run(node_total)
    total_time = TASKS_WITH_DURATIONS_SECONDS.sum(&:duration).to_f
    ideal_time_per_node = total_time / node_total
    tasks_by_duration_desc = TASKS_WITH_DURATIONS_SECONDS.sort_by { |a| -a.duration }
    nodes = Array.new(node_total) { |i| NodeAssignment.new(i + 1, []) }

    puts "Total expected time: #{total_time}; ideal time per job: #{ideal_time_per_node}.\n\n"
    puts "Tasks to distribute:"
    tasks_by_duration_desc.each { |task| puts "* #{task.cmd} (#{task.duration}s)" }

    # Distribute tasks optimally first
    puts "\nAssigning tasks optimally."
    distribute_tasks(tasks_by_duration_desc, nodes, ideal_time_per_node: ideal_time_per_node)

    # Distribute remaining tasks, ordered by ascending duration
    leftover_tasks = tasks_by_duration_desc - nodes.flat_map(&:tasks)

    if leftover_tasks.any?
      puts "\n\nAssigning remaining tasks: #{leftover_tasks.flat_map(&:cmd)}"
      distribute_tasks(leftover_tasks, nodes.sort_by { |node| node.total_duration })
    end

    nodes.each do |node|
      puts "\nExpected duration for node #{node.index}: #{node.total_duration} seconds"
      node.tasks.each { |task| puts "* #{task.cmd} (#{task.duration}s)" }
    end

    nodes
  end

  def distribute_tasks(tasks, nodes, ideal_time_per_node: nil)
    condition =
      if ideal_time_per_node
        ->(task, node, ideal_time_per_node) { (task.duration + node.total_duration) <= ideal_time_per_node }
      else
        ->(*) { true }
      end

    tasks.each do |task|
      nodes.each do |node|
        if condition.call(task, node, ideal_time_per_node)
          assign_task_to_node(tasks, node, task)
          break
        end
      end
    end
  end

  def assign_task_to_node(remaining_tasks, node, task)
    node.tasks << task
    puts "Assigning #{task.command} (#{task.duration}s) to node ##{node.index}. Node total duration: #{node.total_duration}s."
  end
end

if $PROGRAM_NAME == __FILE__
  options = {}

  if ARGV.include?('--dry-run')
    options[:dry_run] = true
  end

  StaticAnalysis.new.run_tasks!(options)
end