Untitled

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

class Article do
  attr_reader :url, :title, :content, :blog, :author, :published, :tags, :comments, :favorites, :rating

  def initialize options
    @url       = options[:url]
    @title     = options[:title]
    @content   = options[:content]
    @blog      = options[:blog]
    @author    = options[:author]
    @published = options[:published]
    @tags      = options[:tags]
    @comments  = options[:comments]
    @favorites = options[:favorites]
    @rating    = options[:rating]
  end
end

class Habr do
  attr_reader :user, :favorites, :pages

  def initialize(user)
    raise 'User is not set' unless user

    @user = user
  end

  def url
    'http://habrahabr.ru'
  end

  def favorites_url
    "#{url}/users/#{user}/favorites/"
  end

  def pages
    unless @pages
      document = Nokogiri::HTML( open(favorites_url, 'User-Agent' => '') )

      @pages = document.at_css('#nav-pages a:last-child')['href'].match(/(\d+)\/$/)[1].to_i
    else
      @pages
    end
  end

  def favorites
    @favorites ||= (1..pages).inject([]) do |articles, page|
      document = Nokogiri::HTML( open(favorites_url + "page#{page}/", 'User-Agent' => '') )

      document.css('.post').each do |article|
        header = article.css('.title')
        bottom = article.css('.infopanel')

        options = {
          url       : header.at_css('.post_title')['href'], # could it be simply #href here?
          title     : header.at_css('.post_title').text,
          content   : nil,
          blog      : header.at_css('.blog_title').text
          author    : bottom.at_css('.author').text
          published : bottom.at_css('.published').text
          tags      : article.at_css('.tags li').map{ |li| li.text }.join(', ')
          comments  : bottom.at_css('.comments .all').text
          favorites : bottom.at_css('.favs_count').text
          rating    : bottom.at_css('.score').text
        }

        pages << Article.new(options)
      end
    end
  end
end

username             = 'gmile'
habr_url             = "http://habrahabr.ru/users/#{username}/favorites/"
favorites_start_page = Nokogiri::HTML( open(habr_url, 'User-Agent' => '') )
pages                = favorites_start_page.css('ul#nav-pages a:last-child').last['href'].match(/(\d+)\/$/)[1].to_i

puts "Fetching links from #{pages} pages..."


puts links.inspect

x = Nokogiri::HTML::Builder.new(:encoding => 'UTF-8') { |doc|
  doc.html {
    doc.head {
      doc.title "Interesting"
    }

    doc.body {
      links[1..1].each { |link|
        article = Nokogiri::HTML( open(link, 'User-Agent' => '') )

        doc.h2 article.at_css('.title').text.strip
        doc.h4 article.at_css('.author').text.strip
        doc.h4 article.at_css('.published').text.strip
        doc.parent << article.at_css('.content')
      }
    }
  }
}

puts 'HTML generation is done'

PDFKit.new(x.doc.to_s, :page_size => 'A4').to_file('out.pdf')