Print NYT Wordle stats: date, hard mode?, # guesses

print_wordle_stats.rb

require 'net/http'
require 'time'
require 'json'
require 'uri'
require 'pry'

TOKEN = File.read('nyt-s.token').rstrip
API_BASE = 'https://www.nytimes.com/svc/games'

DELAY = 0.2
SECONDS_IN_DAY = 24 * 60 * 60

START_DATE = Time.parse('2025-02-24')
END_DATE = Time.parse('2025-03-09')

def format_date(date)
  date.strftime('%Y-%m-%d')
end

def each_puzzle(start_date, end_date, &block)
  uri1 = URI.parse("#{API_BASE}/v1/archive/wordle/#{format_date(start_date)}/#{format_date(end_date)}")
  http = Net::HTTP.new(uri1.host, uri1.port)
  http.use_ssl = true

  request1 = Net::HTTP::Get.new(uri1.request_uri)
  request1['nyt-s'] = TOKEN

  id_map = Hash[JSON.parse(http.request(request1).body).map { |el| [el['id'], el['print_date']] }]

  # result_count = id_map.keys.count
  # expected_count = ((end_date - start_date) / SECONDS_IN_DAY).floor + 1
  # fail "Expected #{expected_count}, got #{result_count}" unless expected_count == result_count

  uri2 = URI.parse("#{API_BASE}/state/wordleV2/latests")
  args = {
    puzzle_ids: id_map.keys.join(',')
  }
  uri2.query = URI.encode_www_form(args)

  request2 = Net::HTTP::Get.new(uri2.request_uri)
  request2['nyt-s'] = TOKEN

  JSON.parse(http.request(request2).body)['states'].sort_by { |el| el['print_date'] }.each do |r|
    next if Time.parse(r['print_date']) < start_date
    next if Time.parse(r['print_date']) > end_date

    print_date = r['print_date'] || ''
    print_date = id_map[r['puzzle_id'].to_i] if print_date.empty?
    yield [print_date, r]
  end
end

d = START_DATE
while d <= END_DATE
  # Maximum number of days allowed in request = 31
  each_puzzle(d, d + 30 * SECONDS_IN_DAY) do |print_date, r|
    game_data = r['game_data']
    next unless game_data['status'] == 'WIN'
    puts [print_date, r['puzzle_id'], game_data['hardMode'], game_data['currentRowIndex']].join("\t")
  end

  d += (31 * SECONDS_IN_DAY)
  sleep DELAY
end