http://twelve-factor-ja.herokuapp.com/ をepub/mobi化するスクリプト

twelve_factor.rb

# -*- coding: utf-8 -*-
require 'nokogiri'
require 'open-uri'
require "FileUtils"

target = "http://twelve-factor-ja.herokuapp.com"

html = Nokogiri::HTML(open(target))

top_article = html.css(".abstract")

url_list = html.css(".concrete h2 > a").collect{|a| target +  a["href"]}
 
article_list = url_list.collect{|url|
   article_doc =  Nokogiri::HTML(open(url))
   article_doc.css(".abstract").first
}

article_list.each{ |article_doc|
  article_doc.css("img").each{ |ele|
    path =target + ele["src"]
    
    fileName = File.basename(path)

    dir = "img/"
    filepath = dir + fileName

    FileUtils.mkdir_p(dir) unless FileTest.exist?(dir)
    
    open(filepath, 'wb') do |output|    
      open(path) do |data| 
        output.write(data.read)       
      end
    end

    ele["src"] = filepath
  }
}

open("temp.html", "w"){ |f|
  f.puts(top_article)
  article_list.each {|article|
    f.puts(article)
  }
}

title = html.title;
author = "Adam Wiggins"

open("temp_title.txt", "w"){ |f|
  f.puts("% #{title}")
  f.puts("% #{author}")
}

`pandoc temp.html -o "#{title}.md"`
`pandoc temp_title.txt "#{title}.md" -o "#{title}.epub"`
`ebook-convert "#{title}.epub" "#{title}.mobi"`