どどんとふのチャットログからダイスロール部分を抽出するスクリプト

txt_to_yaml.rb

require 'yaml'
require 'json'

lines = ARGF.readlines(chomp: true)
json_lines = lines.map { |line|
  left_bracket_index = line.index('[')
  line[left_bracket_index..-1]
}
hashes = json_lines.map { |line|
  JSON.parse(line)
}
messages = hashes.map{ |hash|
  hash[1]['message']
}

sharps = messages.grep(/\A###/)
inner_hashes = sharps.map { |message|
  JSON.parse(message.sub(/\A###.+?###/, ''))
}
chat_messages = inner_hashes.map{ |hash|
  hash['chatMessage']
}.compact
multi_lines = chat_messages.select { |m|
  m.chomp.lines.length > 1
}
dice_commands = multi_lines.
  map { |m|
    second_line = m.lines[1]
    matches = second_line.match(/\A(.+?):/)
    game_id = matches ? matches[1].strip : nil
    [game_id, m.lines[0].strip]
  }.
  reject { |game_id, _| game_id.nil? }.
  sort_by { |game_id, _| game_id }

game_id_messages = dice_commands.reduce({}) { |acc, (game_id, message)|
  acc[game_id] ||= []
  acc[game_id] << message

  acc
}

YAML.dump(game_id_messages, $stdout)