gistfile1.scala

  def annotate(text: String, field: String): String = {
    def prependHttp(s: String): String = if (!(s.startsWith("http://") || s.startsWith("https://"))) "http://" + s else s

    {
      val preprocessed = text.replaceAll(System.getProperty("line.separator"), " ")
      pattern.findAllMatchIn(preprocessed).toList.map(m => {
        Option(m.subgroups(0)).flatMap(token => {
          symbolCode.get(token.slice(0, 1)).map(sym => Seq(sym, field, m.start(1).toString, m.end(1).toString, token, token.substring(1)))
        }).getOrElse(Seq.empty[String]).mkString("|")
      }) ++ urls.findAllMatchIn(preprocessed).toList.filterNot(m => {
        val start: Int = m.start(1)
        val end: Int = m.start(1)
        val isHref: Boolean = (start - 7 > 0) && text.substring(start - 7, start).contains("href")
        val isAnchor: Boolean = (start - 1 > 0 && end + 1 < text.length && text.charAt(start - 1).toString == ">" && text.charAt(end + 1).toString == "<")
        isHref || isAnchor
      }).map(m => {
        Option(m.group(1)).map(token => Seq("l", field, m.start(1).toString, m.end(1).toString, token, prependHttp(token))).getOrElse(Seq()).mkString("|")
      })
    }.mkString(";")
  }