ブサ彦物語ダウンロードする

bigaku.rb

# -*- coding: utf-8 -*-

require 'open-uri'
require 'nokogiri'
require 'uri'

$index_url = URI.parse('http://mbbs.tv/u/?id=18over')
$known_pages = []

def page(url)
  Nokogiri open(url).read.force_encoding('sjis')
end

def abs(url)
  ($index_url + url).to_s
end

def write_file(title, body)
  open("#{title}.txt", "w") {|f|
    f.write body
  }
end

def topic_page(url)
  container = page(url).at('div[align="LEFT"]')
  body = container.text.gsub("[←前][次→] back", "").gsub(/\r/, '').gsub(/^\n+/, '')
  title = body.lines.first.chomp
  p title
  write_file(title, body)
end

def index_page(url)
  return if $known_pages.include? url
  $known_pages << url
  page(url).search('a').each{|a|
    case a['href']
    when /p=\d+/
      index_page(abs(a['href']))
    when /tid=\d+/
      topic_page(abs(a['href']))
    end
  }
end

index_page($index_url)