tweet.rb

# 面接用
require 'cgi'
require 'uri'
require 'strscan'

class TextStyle
  def initialize(rules = nil)
    @rules = Array.new
    if (rules)
      add(rules)
    end
  end
  
  def clear
    @rules.clear
  end
  
  def add(*args)
    if ( args[0].instance_of?(Hash) )
      add_hash(args[0])
    elsif ( args[0].instance_of?(Array) )
      add_hash_array(args[0])
    elsif ( args.size > 1 )
      add_param(args[0], args[1], args.size > 2 ? args[2] : nil)
    else
      raise ArgumentError
    end
  end
  
  def add_hash_array(rules)
    if (rules)
      rules.each do |rule|
        add_hash(rule)
      end
    end
  end
  
  def add_hash(param)
    add_param(param[:re], param[:filter], param[:priority])
  end
  
  def add_param(re, filter, priority = nil)
    if (priority.nil?)
      priority = @rules.empty? ? 0 : @rules[-1][:priority] + 1
    end
    @rules << { :priority => priority, :re => re, :filter => filter }
  end
  
  def eval(text)
    rules = @rules.sort do |a, b|
      a[:priority] <=> b[:priority]
    end
    return eval_rule(text, rules, 0)
  end
  
  protected
  def eval_rule(text, rules, i)
    output = String.new
    
    if (i >= rules.size || text.nil?)
      return output
    end
    scanner = StringScanner.new(text)
    until scanner.eos?
      token = scanner.scan_until(rules[i][:re])
      if (token)
        output << eval_rule(scanner.pre_match, rules, i + 1) + rules[i][:filter].call(scanner.matched)
        scanner.string = scanner.post_match
      else
        output << eval_rule(scanner.string, rules, i + 1)
        break
      end
    end
    return output
  end
end

rules = [{
           :priority => 1,
           :re => /s?https?:\/\/[-_\da-zA-Z\.!~*'\(\);\/?:@&=+$,%\#]+/,
           :filter => Proc.new do |s|
             sprintf('<a class="tweet-url web" target="_blank" rel="nofollow" href="%s">%s</a>',
                     URI.escape(s), CGI.escapeHTML(s))
           end
         },
         {
           :priority => 2,      
           :re => /@[a-zA-Z_]+/,
           :filter => Proc.new do |s|
             sprintf('<a class="tweet-url username" href="/%s">%s</a>',
                     URI.escape(s), CGI.escapeHTML(s))
           end
         },
         {
           :priority => 3,
           :re => /#[a-zA-Z_]+/,
           :filter => Proc.new do |s|
             sprintf('<a class="tweet-url hashtag" href="/search?q=%s">%s</a>',
                     URI.escape(s), CGI.escapeHTML(s))
           end
         },
         {
           :priority => 4,      
           :re => /.+/,
           :filter => Proc.new do |s|
             CGI.escapeHTML(s)
           end
         }
]

s = TextStyle.new(rules)
puts s.eval("@kusobot <s>きた～ http://www.example.net/@kusobot#kusobot o-('-'o)#kusobot hoge")
# <a class="tweet-url username" href="/@kusobot">@kusobot</a> &lt;s&gt;きた～ <a class="tweet-url web" target="_blank" rel="nofollow" href="http://www.example.net/@kusobot%23kusobot">http://www.example.net/@kusobot#kusobot</a> o-('-'o)<a class="tweet-url hashtag" href="/search?q=%23kusobot">#kusobot</a> hoge

rules[0][:priority] = 2
rules[1][:priority] = 1
s.clear
s.add(rules)
puts s.eval("@kusobot <s>きた～ http://www.example.net/@kusobot#kusobot o-('-'o)#kusobot hoge")
# <a class="tweet-url username" href="/@kusobot">@kusobot</a> &lt;s&gt;きた～ <a class="tweet-url web" target="_blank" rel="nofollow" href="http://www.example.net/">http://www.example.net/</a><a class="tweet-url username" href="/@kusobot">@kusobot</a><a class="tweet-url hashtag" href="/search?q=%23kusobot">#kusobot</a> o-('-'o)<a class="tweet-url hashtag" href="/search?q=%23kusobot">#kusobot</a> hoge

s.clear
s.add(rules[1])
s.add(rules[2])
s.add(rules[3])
puts s.eval("@kusobot <s>きた～ http://www.example.net/@kusobot#kusobot o-('-'o)#kusobot hoge")
# <a class="tweet-url username" href="/@kusobot">@kusobot</a> &lt;s&gt;きた～ http://www.example.net/<a class="tweet-url username" href="/@kusobot">@kusobot</a><a class="tweet-url hashtag" href="/search?q=%23kusobot">#kusobot</a> o-('-'o)<a class="tweet-url hashtag" href="/search?q=%23kusobot">#kusobot</a> hoge