TwitterRetweetCrawler.scala

import java.net.{Authenticator,  PasswordAuthentication}
import scala.Math._
import scala.xml._
import scala.io.Source

object TwitterRetweetCrawler {

  val rtUrl ="http://api.twitter.com/statuses/retweets_of_me.xml"
  val stUrl ="http://api.twitter.com/1/statuses/retweets/%s.xml"

  // Usage : scala TwitterRetweetCrawler <your_id> <your_password>
  def main(args:Array[String]) = {
    val Array( userid, passwd ) = args

    // Basic認証
    Authenticator.setDefault(
      new Authenticator {
        override def getPasswordAuthentication =
          new PasswordAuthentication( userid,  passwd.toCharArray)
      }
    )

    // RetweetされたPostを取得する
    val source = Source.fromURL( rtUrl )
    val xml = XML.loadString( source.getLines.mkString )
    (xml \ "status").map{ s =>
      val id = ( s \ "id" text).toLong
      val text = s \ "text" text
      val retweetedBy = {
        val stxml = XML.loadString(
          Source.fromURL( stUrl format id ).getLines.mkString )
        (stxml \ "status" \ "user" ).map { st =>
          ( st \ "name" text , st \ "screen_name" text ) }
      }
      ( id, text, retweetedBy )
    }.foreach{ case( id, text, retweetedBy ) =>
      val cxml = XML.loadString("<c>" + (text) + "</c>")
      println("""
%s
  retweeted by %s users.[%s]
""".format( cxml.child.text, retweetedBy.size, retweetedBy.map{ case(name, screen) =>
     "%s(%s)".format( name, screen ) }.mkString(",")))
    }
  }
}