html_book_to_pdf.rb

#!/usr/bin/env ruby

# Script to convert the Learn Prolog Now! online HTML edition to a PDF.

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

class HTMLBook
  attr_reader :content, :footer
  def initialize(root)
    @root = root + "/"
    @content = ""
    @crawled_pages = []
  end

  def parse_url(url)
    File.basename(url).gsub(/#.*$/, '')
  end

  def add_and_crawl(page)
    doc = Nokogiri::HTML(open(@root + page))
    add page, doc.dup

    doc.css("ul li").css("a").each do |link|
      add link.attr("href")
    end
  end

  def add(page, doc = nil)
    html_base_file = parse_url(page)
    return if @crawled_pages.include? html_base_file
    @crawled_pages << html_base_file

    doc ||= Nokogiri::HTML(open(@root + page))

    # Remove next/up/prev links.
    doc.css("table.nav").each(&:remove)

    # Process footer.
    doc.css("address").each(&:remove)

    # Remove the extra horizantal rule.
    doc.css("hr").last.remove

    # Convert anchor link targets to local targets.
    doc.css("a").each do |link|
      if link.attr("href")
        next if link.attr("href").index("http://")
        link["href"] = "#" + link.attr("href").gsub(/^.*?#/, '')
      elsif link.attr("name")
        link["href"] = "#" + link.attr("name")
      end
    end

    # Append the content.
    @content += "<div id='#{html_base_file}'>"
    @content += doc.css("body").inner_html #.gsub(/[a-zA-Z0-9\.html#/, '')
    @content += "</div>"
  end
end

class Generator
  def initialize(book)
    kit = PDFKit.new(book.content)
    kit.to_file("learn_prolog_now.pdf")
  end
end

book = HTMLBook.new("http://cs.union.edu/~striegnk/learn-prolog-now/html")
book.add_and_crawl("toc.html")

Generator.new(book)