ruby rssserver

require ‘cgi’
require ‘rss’
require ‘net/http’

def parse(page_source)
dates = page_source.scan(%r!(\d+)年 ?(\d+)月 ?(\d+)日
!).map { |year, month, day| [year.to_i, month.to_i, day.to_i] }
url_titles = page_source.scan(%r!^(.+?)
!).map { |url, title| [CGI.unescapeHTML(url), CGI.unescapeHTML(title)] }
url_titles.zip(dates).map { |(url, title), date| [url, title, Time.local(*date)] }
end

def format_text(title, url, url_title_time_ary)
text = “Title: #{title}\nURL: #{url}\n\n”
url_title_time_ary.each do |aurl, atitle, atime|
text << “* (#{atime})#{atitle}\n”
text << ” #{aurl}\n”
end
text
end

def format_rss(title, url, url_title_time_ary)
RSS::Maker.make(“2.0”) do |maker|
maker.channel.updated = Time.now.to_s
maker.channel.link = url
maker.channel.title = title
maker.channel.description = title
url_title_time_ary.each do |aurl, atitle, atime|
maker.items.new_item do |item|
item.link = aurl
item.title = atitle
item.updated = atime
item.description = atitle
end
end
end
end

uri = URI(“http://crawler.sbcr.jp/samplepage.html”)
response = Net::HTTP.get_response(uri)
parsed = parse(response.body.force_encoding(“UTF-8”))

formatter = case ARGV.first
when “rss-output”
:format_rss
else
:format_text
end

puts send(formatter, “WWW.SBCR.JP トピックス”, “http://crawler.sbcr.jp/samplepage.html”, parsed)

投稿者: chosuke

趣味はゲームやアニメや漫画などです

コメントを残す

メールアドレスが公開されることはありません。 が付いている欄は必須項目です