Fast500Parser.scala

import collection.mutable.{ListBuffer, LinkedList}
import io.Source
import scala.collection.mutable.HashMap

/**
 * Date: Apr 16, 2010
 * Time: 10:44:22 PM
 */

object Parser
{
  def main(args: Array[String]) {
    val attributes = List("company", "url", "country", "rate", "field")
    //val all_fields = List("Software", "Telecommunications/Networking", "Media/Entertainment", "Internet", "Biotech/Pharmaceutical/ Medical Equipment", "Other", "Greentech", "internet", "Semiconductor, Components and Electronics", "Computers/Peripherals")
    //val all_countries = List("Turkey", "United Kingdom", "Belgium", "Germany", "Netherlands", "Republic of Ireland", "Poland", "France", "Norway", "Israel", "Northern Ireland", "Sweden", "Russia", "Romania", "Finland", "Bulgaria", "Hungary", "Greece", "Serbia", "Slovakia", "Denmark", "Austria", "Czech Republic", "Portugal", "Croatia", "Estonia")
    import scala.collection.immutable.Set
    val my_fields = Set("Software", "Internet")
    val my_countries = Set("United Kingdom")
    val all_attributes = List("position") ::: attributes
    val heading = "|_. %s |".format(all_attributes.map(_.toUpperCase).mkString(" |_. "))
    //    val all_companies = List[Int]()
    val dir = new java.io.File("/Users/xan/IdeaProjects/ScalaTest/data")
            .listFiles().filter(_.getName().matches("Fast500_\\d+.txt"))
    val company_entry = """(\d+)\s(.*)""".r
    val all_companies = ListBuffer[HashMap[String, String]]()
    for{file <- dir}
      {
        val lines = Source.fromFile(file).getLines().buffered
        val companies = ListBuffer[HashMap[String, String]]()

        while (lines.head.matches(company_entry.pattern.toString)) {
          val company = HashMap[String, String]()
          val company_entry(position, name) = lines.next
          company(all_attributes(0)) = position
          company(all_attributes(1)) = name
          companies.append(company)
        }

        for (attribute <- attributes.tail; company <- companies) {
          company(attribute) = lines.next()
        }
//        println(companies.size)
        all_companies.appendAll(companies)
      }

    all_companies.foreach(
      (m) => print("%s\n".format(
        m.map(
          (i) => "%s\n".format(i.toString)))))
    println("SIZE: %s".format(all_companies.size))
  }
}