delicious-linkchecker.rb

# Configuration:
#   change 'username'
#   change 'max_pages'
#     it's the number of pages of bookmarks 
#     you need to go to delicious and find the exact number manually
#   see Watir::Browser.xxx for extra config parameters
# 
# This code:
#   1- open a browser to your delicious page
#   2- fetch each link
#   3- only keep bookmarks
#   4- check each bookmark http status code
#   5- if status code is wrong: display link in error and open a browser windo to it
#   6- fetc next page and go on
#   ...
#   999- you need to manually delete the failling link
# 
# Watir installation: http://wiki.openqa.org/display/WTR/Installation
require 'watir'
require 'net/http'

username= 'xxx'
max_pages = 3

def display_error(link, msg)
  puts "*** #{link} *** \t\t => \t #{msg}"
end

# open a browser with the error page to viculy checkit
def error_and_browser(link, msg)
  display_error(link, msg)
  Watir::Browser.start link
end

# Watir::Browser.default = "ie" # ie , firefox , safari
# Watir::Browser.speed = 'zippy' # "fast", "slow", "zippy"
# Watir::Browser.visible = true

b = Watir::Browser.new

# look in each page
(1..max_pages).each do |page|
  puts "\n\t\t\t=== page: #{page} of #{max_pages}  ==="
  b.goto("http://delicious.com/#{username}?page=#{page}")
  
  # only keep http:// links
  hrefs = b.links.collect {|link| link.href if link.href =~ /http/}
  # remove username links
  hrefs = hrefs.compact.collect {|href| href unless href.include? username}
  # remove all delicious and yahoo links
  links = hrefs.compact.collect {|href| href unless href =~ /(delicious.com)|(info.yahoo.com)/}
  links.compact.each do |link|
    begin
      response = Net::HTTP.get_response URI.parse(link)
      case response
        when Net::HTTPSuccess then puts "\tok: #{link}"
        when Net::HTTPFound then puts "\tok (302): #{link}"
      else
        error_and_browser(link, "#{response.code} #{response.message}")
      end
    rescue Errno::ETIMEDOUT => detail
      display_error(link, detail)
    rescue Net::HTTPBadResponse => detail
      display_error(link, detail)
    rescue EOFError => detail
      display_error(link, detail)
    rescue Timeout::Error => detail
      display_error(link, detail)
    rescue Errno::ECONNRESET  => detail
      display_error(link, detail)
    end
  end
end