Dedupe placemarks in KML file

kml_dedupe.scala

// Run with "scala convert.scala"
import scala.xml.XML
import scala.xml.Node

// Given a node, return a trimmed string containing its coordinates
def toCoords(node: Node): String = node match {
  case node if ! (node \ "Point").isEmpty => (node \ "Point" \ "coordinates").text.trim
  case node if ! (node \ "LineString").isEmpty => (node \ "LineString" \ "coordinates").text.trim
  case node if ! (node \ "Polygon").isEmpty => (node \ "Polygon" \ "outerBoundaryIs" \ "LinearRing" \ "coordinates").text.trim
  case _ => println(f"node unknown coords: %s", node \ "name"); "unknown"
} 

// Given two nodes, try to pick the best one.
// TODO: Actually merge the descriptions rather than picking biggest
def mergeNodes(a: Node, b: Node): Node = {
  val firstDesc = (a \ "Description").text.trim
  val secondDesc = (b \ "Description").text.trim
  if (firstDesc.length > secondDesc.length) a
  else b
}

val doc = XML.loadFile("doc.kml")

// val placenodes = doc \ "Folder" \ "Document" \ "Folder" \ "Folder" \ "Folder" \ "Placemark"
val placenodes = doc \\ "Placemark"

var places = placenodes.foldLeft(Map[String,Node]()) {
  (acc, n) =>
    val coord = toCoords(n)
    if (acc.contains(coord)) acc + (coord -> mergeNodes(acc(coord), n))
    else acc + (coord -> n) 
}

places.values.foreach(println)