Ruby strings that do web things

earl.rb

# Earl: Strings that do web things
# 
# 
# Examples:
# 
# text = "Blah blah http://tinyurl.com/4bnjzbu blah http://tinyurl.com/4tefu9f"
# 
# text.urls # => ["http://tinyurl.com/4bnjzbu", "http://tinyurl.com/4tefu9f"]
# 
# text.locations # => ["http://nutrun.com/weblog/2010/11/17/supercharged-ruby-console-output.html", "http://nutrun.com"]
# 
# text.linkify # => "Blah blah <a href=\"http://tinyurl.com/4bnjzbu\">http://tinyurl.com/4bnjzbu</a> blah <a href=\"http://tinyurl.com/4tefu9f\">http://tinyurl.com/4tefu9f</a>"
# 
# "http://tinyurl.com/4bnjzbu".location # => "http://nutrun.com/weblog/2010/11/17/supercharged-ruby-console-output.html"
# 
# puts "http://nutrun.com/".get
# 
# 
# Public domain. Contact twitter.com/nutrun
# -

require 'uri'
require 'net/http'

class String
  def urls
    URI.extract(self, ['http', 'https'])
  end
  
  def location
    final_url_destination(self)
  end
  
  def locations
    self.urls.map { |url| url.location }
  end
  
  def linkify
    self.urls.inject(self) { |s, u| s.gsub(u, %Q{<a href="#{u}">#{u}</a>}) }
  end
  
  def linkify!
    self.urls.each { |u| self.gsub!(url, %Q{<a href="#{u}">#{u}</a>}) }
  end
  
  def get
    uri = URI.parse(self)
    path = uri.path
    path = '/' if path.empty?
    path = "#{path}?#{uri.query}" if uri.query.any?
    Net::HTTP.start(uri.host, uri.port) { |http| http.get(path) }.body
  end
  
  private
  
  def final_url_destination(url, limit=10)
    raise ArgumentError, 'HTTP redirect too deep' if limit == 0
    uri = URI.parse(url)
    path = uri.path
    path = '/' if path.empty?
    path = "#{path}?#{uri.query}" if uri.query
    response = nil
    Net::HTTP.start(uri.host, uri.port) {|http| response = http.head(path) }
    
    case response
      when Net::HTTPSuccess
        url
      when Net::HTTPRedirection
        final_url_destination(response['location'], limit - 1)
      else
        STDERR.puts url
        response.error!
    end
  end
end