audit the privacy settings on documents uploaded to the scribd api

gistfile1.rb

# This script will verify the privacy settings on documents you have loaded
# to the scribd api.

require 'rubygems'
require 'hpricot'
require 'open-uri'


def get_docs(api_uri, limit, offset)
  api_uri += "&limit=#{limit}&offset=#{offset}"
  puts "request api with limit:#{limit} and offset:#{offset}"
  puts api_uri
  Hpricot(open(api_uri))
end

def audit_privacy_settings(api_key)
  limit = 1000
  offset = 0
  public_docs = []
  count = 0
  api_uri = "http://api.scribd.com/api?api_key=#{api_key}&method=docs.getList&use_api_account=1"


  results = (get_docs(api_uri, limit, offset)/"result")
  while results.size > 0
    results.each do |r|
      unless (r/"secret_password").size > 0
        public_docs.push(r)
        puts "public document found: total is #{public_docs.size}"
      else
        puts((r/"doc_id").inner_html + " is private")
      end
      count += 1
    end
    offset +=1000
    results = (get_docs(api_uri, limit, offset)/"result")
  end

  puts "*"*30
  puts "PUBLIC DOCUMENTS"
  puts "*"*30

  public_docs.each do |d|
    puts((d/"doc_id").inner_html + " is a public document!")
  end
  
  puts "*"*30
  puts "SUMMARY"
  puts "*"*30
  puts "total documents: #{count}; public_documents: #{public_docs.size}; private documents : #{count - public_docs.size}"

end

if __FILE__ == $0
  if ARGV[0]
    audit_privacy_settings(ARGV[0])
  else 
    puts "usage: please provide your api key"
  end
end