aturl.rb

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

def get_doc(uri)

	if uri.to_s =~ /explore\.twitter\.com/
		uri = URI.parse('http://twitter.com/' + uri.to_s.scan(/.+\/(.+)$/)[0][0])
	end

	begin
		page = open(uri.to_s).read
	rescue Exception
		return nil
	end
	
	if page == ''
		return nil
	end

	doc = Hpricot.parse(page)
	doc.search('a').each do |a|
		a_uri = URI.parse(a.attributes['href'].gsub(' ','%20')).normalize rescue URI.parse('')
		a_uri.scheme = uri.scheme if a_uri.scheme.nil?
		if a_uri.scheme =~ /https?|ftp/
			a_uri.host = uri.host if a_uri.host.nil?
			a_uri.path = "#{uri.path}#{a_uri.path}" if a_uri.path.nil? || a_uri.path[0..0] != '/'
		end
		a.set_attribute 'href', a_uri.to_s
	end
	doc.search('img').each do |a|
		a_uri = URI.parse(a.attributes['src'].gsub(' ','%20')).normalize rescue URI.parse('')
		a_uri.scheme = uri.scheme if a_uri.scheme.nil?
		if a_uri.scheme =~ /https?|ftp/
			a_uri.host = uri.host if a_uri.host.nil?
			a_uri.path = "#{uri.path}#{a_uri.path}" if a_uri.path.nil? || a_uri.path[0..0] != '/'
		end
		a.set_attribute 'src', a_uri.to_s
	end

	doc

end

def representative_hcard(doc, url)

	uri = URI.parse(url.strip).normalize
	url = uri.to_s

	cards = doc.search('.vcard')
	return cards[0] if cards.size == 1

	cards.each do |vcard|
		uid = vcard.at('.uid')
		unless uid.nil?
			begin
				if uid.attributes['href'].reverse[0..0] == '/' && url.reverse[0..0] != '/'
					iz = (uid.name == 'a' && uid.attributes['href'] == "#{url}/")
				else
					iz = (uid.name == 'a' && uid.attributes['href'] == url)
				end
				if iz || (URI.parse(uid.text_content).normalize.to_s == url)
					return vcard
				end
			rescue Exception
			end
		end
		uid = vcard.at('a[@rel~=me]')
		unless uid.nil?
			return vcard
		end
	end

	nil

end

doc = get_doc(URI.parse(ARGV[0]))
hcard = representative_hcard(doc, ARGV[0])
nick = doc.at('.screen-name') unless hcard
nick = hcard.at('.nickname') unless nick
nick = hcard.at('.fn') unless nick
nick = nick.html.gsub(/<[^>]+>/,'').gsub(/[^a-zA-Z0-9_\-]+/,'.')

latest = doc.at('*[@rel~=bookmark"]').attributes['href'] rescue nil
latest = ARGV[0] unless latest

puts '<span class="vcard reply"> @<a href="' + CGI::escapeHTML(ARGV[0]) + '" class="fn nickname url">' + CGI::escapeHTML(nick) + '</a> </span>'
puts '<a rel="reply" href="' + CGI::escapeHTML(latest) + '">in reply to...</a>' if latest