exercism-comments.rb

require 'json'
require 'time'

# Create a personal access token at https://github.com/settings/tokens
# It needs NO scopes.
# In fact, you could just not use a token, but then you get rate-limited to 60 an hour instead of 5000 an hour.
TOKEN_FILE = 'secrets/token.txt'.freeze
SLEEP_TIME = 2

CACHE = 'cache'.freeze

Dir.mkdir(CACHE) unless File.exist?(CACHE)

LAST_REQUEST = [0]

def req(url)
  cache = "#{CACHE}/#{url.gsub(/\W/, ?_)}"
  if File.exist?(cache)
    res = File.read(cache)
  else
    if Time.now.to_i - LAST_REQUEST.first < SLEEP_TIME
      puts "#{url}: sleeping to not overload."
      Kernel.sleep(SLEEP_TIME)
    end
    LAST_REQUEST[0] = Time.now.to_i
    tok = File.read(TOKEN_FILE).chomp
    res = `curl -H "Authorization: token #{tok}" "#{url}"`
    File.open(cache, ?w) { |f| f.puts(res) }
  end
  JSON.parse(res)
rescue => e
  puts res
  puts "No parse: #{e}"
end

REPO = ARGV.first

case REPO
when 'xgo'
  COMMENT_PAGES = 14
  EVENT_PAGES = 31
  ISSUE_PAGES = 7
when 'xrust'
  COMMENT_PAGES = 11
  EVENT_PAGES = 15
  ISSUE_PAGES = 3
when 'xhaskell'
  COMMENT_PAGES = 12
  EVENT_PAGES = 27
  ISSUE_PAGES = 6
else raise "Unknown repo #{REPO}"
end

comments = (1..COMMENT_PAGES).flat_map { |page|
  req("https://api.github.com/repos/exercism/#{REPO}/issues/comments?per_page=100&page=#{page}")
}

events = (1..EVENT_PAGES).flat_map { |page|
  req("https://api.github.com/repos/exercism/#{REPO}/issues/events?per_page=100&page=#{page}").select { |i| %w(closed reopened).include?(i['event']) }
}

issues = (1..ISSUE_PAGES).flat_map { |page|
  req("https://api.github.com/repos/exercism/#{REPO}/issues?per_page=100&state=all&page=#{page}").reject { |i| i['pull_request'] }
}.map { |i| [i['url'], i] }.to_h

issues.each_value { |i|
  i[:events] = [
    # Probot uses 'updated:' search, and this *DOES* take updated_at into account.
    # Unfortunately, we only get the most recent update.
    # So for issues that were edited after they were closed, we may have false positives.
    # (we think it's stale, when it's was not at the time due to an update)
    [:fresh, Time.parse(i['updated_at'])],
  ]
}

comments.each { |c|
  next unless (i = issues[c['issue_url']])
  i[:events] << [:fresh, Time.parse(c['created_at'])]
  # Probot uses 'updated:' search, and this *DOES* take updated_at into account.
  # Same caveat as above with only getting the most recent update.
  i[:events] << [:fresh, Time.parse(c['updated_at'])]
}

events.each { |e|
  next unless (i = issues[e['issue']['url']])
  event = case e['event']
  when 'closed'; :close
  when 'reopened'; :open
  else raise "unknown event #{e}"
  end
  i[:events] << [event, Time.parse(e['created_at'])]
}

DAY = 86400
STALE = 60 * DAY

issues.each_value { |i|
  open = true
  time = Time.parse(i['created_at'])
  stale = false
  check = ->(etime) { stale ||= open && etime - time >= STALE }
  i[:events].sort_by(&:last).each { |event, etime|
    case event
    when :fresh; check[etime]
    when :close; check[etime]; open = false
    when :open; open = true
    else raise "Unknown event #{event}"
    end
    time = etime
  }
  check[Time.now]
  next unless stale
  puts "%3d %s" % i.values_at('number', 'title')
}