README

Just a simple script to grab AddressBook cards from the social network StudiVZ (which is still lacking an API to do just this)

crawl.rb

require 'rubygems'
require 'studi_crawler'

def getpassword username
  print "Password for "+username+": "
  system "stty -echo"
  pw = gets.delete("\n")
  system "stty echo"
  puts
  pw
end

def getuser 
  print "Username: "
  user = gets.delete("\n")
  puts
  user
end

username = getuser
password = getpassword(username)
StudiCrawler.new(username, password).export_friends "studi"

studi_crawler.rb

#!/usr/bin/env ruby1.9
require 'mechanize'
require 'vcard'

class StudiCrawler

  attr_accessor :login_url
  @login_url = 'https://secure.studivz.net/Login'

  def initialize mail,pw
    @agent = WWW::Mechanize.new
    @page = @agent.get login_url
    login mail,pw
  end

  # tries to fill in the forms and login
  def login mail,pw
    forms = @page.forms.first
    forms.email=mail
    forms.password=pw
    @page = forms.click_button
  end

  # Goes to a page
  # This can be either a full url, a sub-url 
  # (which is then appended) or a friends name,
  # for which an initial url-lookup is done 
  def goto newPage
    unless friends[newPage.to_sym].nil?
      uri = friends[newPage.to_sym].uri
    end
    uri ||= newPage
    uri = @page.uri+uri unless uri.to_s =~ /^http/u
    @page = @agent.get uri
    @page
  end

  # lazy accessor
  def friends
    crawl_friends if @friends.nil?
    @friends
  end

  def friend name
    @friends[name.to_sym]
  end

  # Accessing details. Lazily tries to 
  # retrieve them using the get_DETAIL method
  [:image, :birthday].each do |item|
    define_method(item) do |name|
      f = friend(name)
      if f.send(item).nil?
        curpage = @page
        unless @page.uri.to_s =~ Regexp.new(f.uri.to_s+"$")
          goto f.uri
        end
        send(("get_"+item.to_s).to_sym, f)
        @page = curpage
      end
      f.send(item)
    end
  end

  def get_friends
    @friends ||= {}

    #+TODO : refactor this out!
    afterIdx = @page.links.index(@page.link_with(:text => 'Alle Freunde'))
    curfriends = @page.links[afterIdx+1..-3].select do |l| 
      ((@page.links[@page.links.index(l)+2].text.include? 'Freunde') and (l.uri.to_s =~ /\/Profile\//))
    end

    curfriends.each do |item|
      @friends[item.text.to_sym] = VCard.new
      @friends[item.text.to_sym].uri = item.uri
      @friends[item.text.to_sym].name = item.text
    end
    @friends
  end

  def get_birthday friend
    page = @page.link_with(:text => /^[0-9][0-9]\.[0-9][0-9]\.[1-9][0-9][0-9][0-9]$/u)
    friend.birthday = page.text unless page.nil?
  end

  def extract_image_uri(id)
    img = @page.search(id)
    unless img.nil?
      img.to_a.compact! 
      unless img.first.nil?
	return img.first.attributes['src'] 
      end
    end
    nil
  end

  def get_image(friend)
    friend.image = extract_image_uri("#profileImage")
  end

  def get_next_symbol
    @page.links.select do |l| 
      begin
        l.uri.to_s =~ /\/p\/2$/u
      rescue URI::InvalidURIError
      end
    end.compact.last.text
  end

  def fill_details
    get_friends.each do |name,values|
      oldpage = @page
      goto values.uri
      birthday name
      image name
      print "."
      @page = oldpage
    end
  end

  def crawl_friends
    oldpage = @page
    # make sure we're on the friends list
    goto @page.link_with(:text => 'Meine Freunde').uri
    next_page_sym = get_next_symbol
    fill_details
    while [email protected]_with(:text => next_page_sym).nil?
      puts "Next page!"
      goto @page.link_with(:text => next_page_sym).uri
      fill_details
    end 
    @page = oldpage
  end

  def fill_img_uri
     extract_image_uri("#PhotoContainer")
  end

  def crawl_album link
    oldpage = @page
    goto link.as_uri
    fill_img_uri
    while [email protected]_with(:text => "nächstes Foto >>").nil?
       puts "Next Picture!"
       goto @page.link_with(:text => "nächstes Foto >>").uri
       fill_img_uri
    end
    @page = oldpage
  end

  def export_friends filename
    friends.each do |name,vCard|
      vCard.export(filename)
    end
  end
end

vcard.rb

require 'ostruct'
require 'base64'
require 'curb'

# This is a very simple implementation to 
# create vCards. Details are simply added 
# directly through the OpenStruct interface.
# On exporting the vCard, for each mapping 
# that exists in the class-dictionary the 
# named property in the OpenStruct is checked 
# and a formatted output hopefully adhering to 
# the vCard standard V3 is generated 
class VCard < OpenStruct
  def export filename
    filename += ".vcf"
    File.open(filename, 'a') do |f|
      f << "BEGIN:VCARD\nVERSION:3.0\n"
      @@mapping.each do |field,lambda|
        if respond_to? field
          f << @@mapping[field].call(self.send(field))
        end
      end
      f << "END:VCARD\n"
    end
  end

  @@mapping = {
    :name => lambda { |n| 
      s = n.split
      "N:"+s[1..-1].join(" ")+";"+s.first+";;;\n"+"FN:"+n+"\n" },
    :phone => lambda { |n|
      "TEL:"+n+"\n" },
    :birthday => lambda { |n|
      "BDAY;value=date:"+n.split(".").reverse.join+"\n" },
    :image => lambda { |n|
      img = Base64.encode64(Curl::Easy.perform(n).body_str).gsub(/\s+/s, "\n   ")
      "PHOTO;BASE64:\n   " + img.gsub(/   $/, "")
    }
  }
end