root/lang/ruby/maraigue-twitter/sokuhobot/sokuhobot.rb @ 35316

#!/usr/bin/env ruby
# -*- coding: utf-8 -*-
require "rubygems"
require "json"
require "open-uri"
require "twbot"

require "twsearch"
require "twsearch-def"
require "tinyurl"

require 'net/http'
Net::HTTP.version_1_2

MARKS = '['+Regexp.quote(' ():;<>[]{|}')+']|'+(
        %w'〔 〕 《 》 （ ） ［ ］ ｛ ｝ 〈 〉 「 」 『 』 【 】 ＜ ＞ ≪ ≫ 　 ： ； ｜'.join('|'))
EXCLUDED_USER = %w{
        47news 2NN googlenewsjp headline kagawa_news moviewalker ryukyushimpo
        net_newsj gizmodojapan favoratory j_np bitexblog hrsmtwt
        bombtter bombtter_raw tetsuwo_bot zerotter suzakutter meitantei_bot
        muskabot drizzlebot zetsubot matsuokabot nyabomb now_ hatena_b
}

def tweet2id(url)
        if url =~ /^http:\/\/twitter\.com\/[0-9A-Z_a-z]+\/status(?:es)?\/(\d+)$/
                $1.to_i
        else
                nil
        end
end

class String
        def strip_all
                res = self.dup
                res = $' if res =~ /\A(?:\s|　)+/
                res = $` if res =~ /(?:\s|　)+\z/
                res
        end
end

class SokuhoBot < TwBot
        def load_data
                result = []
                
                @config['search'] ||= {}
                tsu = TwitterSearch::Updater.new(TWS_engines, @config['search'])
                data, conf, error = tsu.search('速報')
                error.each_pair do |name, status|
                        @logmsg += "[Error in loading from search engine: #{name}]\n#{status}\n"
                end
                @config['search'] = conf
                
                # 古いデータの削除
                return [] if data.empty?
                
                data.each do |x|
                        # 先頭が「【速報】」のような形になっているか判定
                        if x["text"] =~ /\A(?:#{MARKS})*速報(?:#{MARKS})/
                                # 有効な速報postであるか判定
                                value = $'.strip_all
                                
                                next if value =~ /\A(?:#{MARKS}|　|\s)+\z/
                                next if EXCLUDED_USER.include?(x['user'])
                                
                                # 発言URLをtinyurlで短縮
                                # (Twitterが自動短縮しようとする場合のみ)
                                permalink = nil
                                begin
                                        # Twitterが自動短縮しない場合はそのまま利用
                                        raise unless x['link'].index("_")
                                        
                                        # tinyurlで短縮
                                        tinyurl_key = TinyURL.convert(x['link'])
                                rescue Exception => e
                                        # tinyurlが利用出来ないか、する必要がなかった場合
                                        permalink = x['link']
                                else
                                        # tinyurlを利用出来た場合
                                        permalink = "http://tinyurl.com/#{tinyurl_key}"
                                end
                                
                                # 発言を生成
                                message = "[#{permalink} #{x['user']}より速報] #{value}"
                                message += " -b" if message.index("爆発しろ")
                                
                                # Reverse order so results are sorted by original order
                                result.unshift TwBot.remove_reply(message)
                        end
                end
                
                @logmsg = "#{result.size} messages added."
                result
        end
end

SELFDIR = File.dirname(__FILE__)
ARGV.each do |mode|
        #SokuhoBot.new mode, "#{SELFDIR}/config.yml", "#{SELFDIR}/error.log", '', false, true
        SokuhoBot.new mode, "#{SELFDIR}/config.yml", "#{SELFDIR}/error.log"
end
if ARGV.empty?
        open "#{SELFDIR}/error.log", "w" do |file|
                STDERR.puts "No mode specified!"
                file.puts "No mode specified!"
        end
end