build.sbt

name := "wot-scala"

version := "0.0.1"

scalaVersion := "2.10.2"

libraryDependencies ++= Seq(
  "org.jsoup" % "jsoup" % "1.7.2",
  "com.fasterxml.jackson.module" %% "jackson-module-scala" % "2.1.3"
)

ParseVehicleData.scala

import java.io.File
import scala.collection.JavaConversions._
import org.jsoup.Jsoup
import org.jsoup.nodes.Element

object ParseVehicleData extends App {
  val baseURL = "http://worldoftanks.com"

  val categories = Map(
    "Light Tanks" -> "lt",
    "Medium Tanks" -> "mt",
    "Heavy Tanks" -> "ht",
    "Tank Destroyers" -> "td",
    "SPGs" -> "spg")

  val tiers = Map(
    "I" -> 1,
    "II" -> 2,
    "III" -> 3,
    "IV" -> 4,
    "V" -> 5,
    "VI" -> 6,
    "VII" -> 7,
    "VIII" -> 8,
    "IX" -> 9,
    "X" -> 10)

  case class Vehicle(id: String, name: String, category: String, tier: Int, url: String)

  def parseVehicle(category: String, elem: Element) = {
    val name = elem.select(".b-encyclopedia-list_name").text
    val relativeURL = elem.select(".b-encyclopedia-list_linc").attr("href")
    val id = relativeURL.split("/").last.toLowerCase
    val tierText = elem.select(".b-encyclopedia-list_level").text
    val tier = tiers(tierText)
    val url = baseURL + relativeURL
    Vehicle(id, name, category, tier, url)
  }

  val input = new File("wot-encyclopedia-20130819.html")
  val doc = Jsoup.parse(input, "UTF-8", "http://worldoftanks.com/encyclopedia/vehicles/")
  doc.select(".b-encyclopedia-type").iterator.foreach { elem =>
    val category = categories(elem.text)
    elem.nextElementSibling.select(".b-encyclopedia-list_point").iterator.foreach { elem =>
      val vehicle = parseVehicle(category, elem)
      println(vehicle)
    }
  }
}