Clean up twitter handles

twitter_handle_cleanup.rb

#!/usr/bin/env ruby

require "rubygems"
require "active_record"
require "uri"
require "yaml"

config_file = File.open(File.expand_path('~') + "/etc/smt_config.yml")
config = YAML.load(config_file)
sites = config["sites"]


sites.each do |s|

  puts "*** #{s[0]}"

  db = "stage_#{s[1]['db']}"

  ActiveRecord::Base.establish_connection(
  :adapter => "mysql",
  :host => "127.0.0.1",
  :port => 3334,
  :username => "",
  :password => "",
  :database => db
  )

  class Profile < ActiveRecord::Base
    set_table_name "content_type_profile"
  end

  tally = Hash.new(0)

  Profile.all.each do |p|

    twitter = p.field_twitter_value
    next if twitter == nil


    begin

      case
      when twitter.downcase.match(/^(www\.|)twitter.com/)
        new_twitter = "@" + twitter.downcase.sub(/.+?witter.com\//, "")
        puts "#{twitter} is in non-URI URL format. Converting to => #{new_twitter}"
        tally["non-URI URL"] +=1
      when twitter.downcase.match(/^http(s:|:).+twitter\.com\/\#\!/)
        new_twitter = "@" + twitter.sub(/^.+?\#\!(.+?)/, "\0").downcase
        puts "#{twitter} is in hashbang format. Converting to => #{new_twitter}"
        tally["hashbang"] +=1
      when twitter.match(/^@\w{1,}/)
        next
        tally["ok"] +=1
        puts "#{twitter} is in @handle format"
      when twitter.downcase.match(/^http.+?twitter.com\/\w+?$/)
        new_twitter = "@" + URI.parse(twitter).path.gsub(/^\//,"").downcase
        puts "#{twitter} is in URL format. Converting to => #{new_twitter}"
        tally["URL"] +=1
      else
        new_twitter = "@#{twitter.downcase}"
        puts "#{twitter} is in non-@ handle format. Converting to => #{new_twitter}"
        tally["non-@ handle"] +=1
      end
    rescue => ex
      puts ex.message
    end

  end



  puts "\n\n\n*** Tally for #{s[0]}"
  tally.each do |t,v|
    puts "#{t}: #{v}"
  end

  puts "\n\n-----------------------------------\n\n"
end