Create JSON of all books from it-ebooks.info.

it-ebooks.rb

#!/usr/bin/env ruby
# -*- encoding: utf-8 -*-
#
# Thu Nov 28 15:47:21 CET 2013
# 2013 (c) René Wilhelm <[email protected]>
#

# jq '.publisher + ": " + .name + " (" + .author + ")"' it-books.json

# Todo: Exit condition
# Implement a way to exit when done, after approx. 1300 repetitions.

require 'fileutils'
require 'json'
require 'mechanize'
require 'nokogiri'
require 'open-uri'
require 'ostruct'
require 'pry'

url = "http://it-ebooks.info/"

def getBook(url) # => JSON
	p = Nokogiri::HTML(open(url))
	a = p.css('*[@itemprop]').collect { |x| [x['itemprop'], x.text] }
  h = a.inject(Hash.new{ |h,k| h[k]="" }){ |h,(k,v)| h[k] = v; h }
	d = p.search('//tr[11]/td[2]/a').first.attributes['href'].text
	h['download'] = d
	h['url'] = url
	OpenStruct.new h
end

def writeJSON(b)
  File.open("it-books.json", "a") do |f|
    f.write(JSON.pretty_generate(b.to_h))
  end
end

def downloadPDF(b)
  a = Mechanize.new
  a.get(b.url) do |p|
    pdf = p.link_with(:href => b.download).click

    path = "#{b.publisher}/"
    filename = "#{b.author.split(",").map {|x| x.split.last.capitalize}.join(" ")} #{b.datePublished} #{b.name}.pdf"

    Dir.exists?(path) || FileUtils.mkdir_p(path)

    File.open(path + filename, "w") do |f|
      f.write(pdf.body)
    end
  end
end

n = 0
begin
  loop do
    n += 1
    b = getBook(url + "book/#{n}")

    writeJSON(b)

    if b.description.downcase =~ ARGV[0] && b.datePublished.to_i >= ARGV[1]
      downloadPDF(b)
      puts "#{b.datePublished} [#{b.publisher}] #{b.name} (#{b.author})"
    end

    printf("Matching books: \t%s\r", n)
    sleep rand(1..(Math::PI))
  end
rescue
  retry
end