Mastodonのトゥートをweathertypingのワードファイルに変換するやつ

weathertyping.rb

require 'net/http'
require 'json'
require 'sanitize'
require 'uri'
require 'nkf'
require 'natto'
require 'jumanpp_ruby'

MASTODON_SERVER = "handon.club"
MASTODON_USER_ID = 1 # highemerly: 1, seibe: 81
DEBUG = true
N_PAGE = 30

HASHTAG_SEPARATORS = "_\u00B7\u200c"
HASHTAG_NAME_RE    = "([[:word:]_][[:word:]#{HASHTAG_SEPARATORS}]*[[:alpha:]#{HASHTAG_SEPARATORS}][[:word:]#{HASHTAG_SEPARATORS}]*[[:word:]_])|([[:word:]_]*[[:alpha:]][[:word:]_]*)"
HASHTAG_RE         = /(?:^|[^\/\)\w])#(#{HASHTAG_NAME_RE})/i
CUSTOMEMOJI_RE     = /(?<=[^[:alnum:]:]|\n|^)
                     :([a-zA-Z0-9_]{2,}):
                      (?=[^[:alnum:]:]|$)/x
EMOJI_RE           = /\p{Emoji}/
KANJI_RE           = /[一-龠々]/

last_id = 200_000_000_000_000_000 # 200000000000000000
count = 0

class String
  def yomi
    yomi = NKF.nkf('-w -X', self)
    yomi.tr('ァ-ン','ぁ-ん')
        .tr('０-９ａ-ｚＡ-Ｚ','0-9a-zA-Z')
        .tr('：；＜＞［］｛｝',':;<>[]{}')
        .tr('？！＄％＃＆＊＠￥','?!$%#&*@¥')
        .gsub(/[[:space:]]/, '')
        .gsub(/(〜|−|—)/, 'ー')
        .gsub(/[★☆※○×→←↑↓]/, '')
        .tr('･', '・')
        .tr('\\', '')
  end
end

class Toot
  def self.format(toot)
    str = Sanitize.clean(toot["content"]).strip.chomp
    str = str.gsub(HASHTAG_RE, '') #hashtag
             .gsub(CUSTOMEMOJI_RE, '') #custom emoji
    URI.extract(str).uniq.each do |url|
      str.gsub!(url, '')
    end
    str
  end

  def self.accept?(toot)
    true
    #toot["mentions"].empty? && toot["visibility"] == "public" && toot["favourites_count"].to_i >= 2
  end
end

(1..N_PAGE).each do
  uri = URI.parse("https://#{MASTODON_SERVER}/api/v1/accounts/#{MASTODON_USER_ID}/statuses")
  # https://#{MASTODON_SERVER}/api/v1/bookmarks
  # https://#{MASTODON_SERVER}/api/v1/favourites"
  uri.query = URI.encode_www_form({ max_id: last_id })

  http = Net::HTTP.new(uri.host, uri.port)
  http.use_ssl = uri.scheme === "https"
  headers = { "Authorization" => "Bearer #{ENV['MASTODON_ACCESS_TOKEN']}" }
  response = http.get(uri, headers)

  JSON.parse(response.body).each do |toot|
    last_id = toot["id"].to_i if last_id > toot["id"].to_i

    if Toot.accept?(toot) then
      count = count + 1
      status = Toot.format(toot)

      if status.length > 0 then
        puts "原文: #{toot["content"]}" if DEBUG
        puts "#{status} (@#{toot["account"]["username"]})"

        JumanppRuby::Juman.new(force_single_path: :true).parse(status) do |word_juman|
          yomi = word_juman[1].to_s.yomi

          if yomi =~ KANJI_RE then # Jumanppでの解析失敗時はMecabに回す
            yomi_katakana = ""
            Natto::MeCab.new('-Oyomi').parse(yomi) do |word_mecab|
              yomi_katakana = yomi_katakana + word_mecab.feature.to_s.chomp
            end
            yomi = yomi_katakana.yomi
          end

          print yomi
        end

        puts ""
      end
    end
  end
end

puts " count= #{count}, max_id= #{last_id}"