Script that downloads all invoices in your KabelBW login center. (Read: I’m not going to cope with KabelBW’s cr… issues)

kabelbw-rechnungen.rb

#!/usr/bin/ruby
# encoding: utf-8

# Public Domain
# (someone claimed the next lines would be useful for… people. So here goes:
# © 2012 Stefan Breunig
# stefan+kabelbwrechnungen@mathphys.fsk.uni-heidelberg.de)

USER = "DEIN KABELBW ACCOUNT"
PASS = "DEIN KABELBW PASSWORT"
PATH = "ABSOLUTER PFAD AN DEM DIE PDFS ABGELEGT WERDEN SOLLEN"

require "rubygems"
require "mechanize"

a = Mechanize.new
puts "Logging in…"
a.get("http://www.kabelbw.de/rechnungen") do |login_page|
  logged_in = login_page.form_with(:name => 'login') do |f|
    f.cscLoginName = USER
    f.cscLoginPwd  = PASS
  end.click_button
  links = logged_in.links_with(:text => "Rechnung")
  pdf_ids = links.map { |l| l.attributes["onclick"] }
  pdf_ids.map! { |id| id.split(",").map { |s| s.match(/'[^']+/).to_s[1..-1] } }
  # selects the dates by going up to the <tr> element, then selecting the
  # first <td>’s text content
  dates = links.map { |l| l.node.parent.parent.children.first.children.text }
  dates.map! { |d| Date.parse(d).strftime }

  puts "Found #{pdf_ids.size} PDFs…"

  pdf_ids.each_with_index do |id, ind|
    if Dir.glob(File.join(PATH, "*#{id[1]}*.pdf")).any?
      puts "Skipping #{id[1]}: Already exists"
      next
    end
    puts "Processing #{id[1]}…"
    pdf = logged_in.form_with(:name => "pdfview") do |f|
      f.cscInvoiceType  = id[0]
      f.cscInvoiceId    = id[1]
      f.cscInvoiceTitle = id[2]
    end.click_button

    filename = File.join(PATH, "#{id[0]}_#{dates[ind]}_#{id[1]}.pdf")
    File.open(filename, 'w+b') { |file| file << pdf.body }
  end

  puts "Logging out…"
  logged_in.form_with(:name => "loginboxform").click_button
end

puts "Thank you, come again."