KML データを JSON に変換するスクリプト

KmlParser.rb

#!ruby
# -*- encoding: UTF-8 -*-
require "rubygems"
require "rexml/document"
require "rexml/streamlistener"
require "sqlite3"
require "pr_geohash"
include REXML

class KmlListener 
  include StreamListener
  @@placemark_context = ['kml', 'Document', 'Folder', 'Placemark']
  @@name_context = @@placemark_context + ['name']
  @@description_context = @@placemark_context + ['description']
  @@coordinate_context = @@placemark_context + ['Point', 'coordinates']

  def initialize
    @context = []
    @array = []
  end

  def array
    @array
  end

  def tag_start(tag, attrs)
    @context.push tag
    case tag
    when "kml"
      @id = 0
    when "Placemark"
      @place = Hash::new
      @id += 1
      @place["id"] = @id
      @place["name"] = ""
      @place["description"] = ""
    end
  end

  def tag_end(tag)
    case tag
    when "Placemark"
#      puts @place  
      @array.push @place
    end
    @context.delete_at -1
  end

  def cdata(content)
    case @context.last
    when "name"
      if @context == @@name_context then
        @place["name"] = content
      end
    when "description"
      if @context == @@description_context then
        /<b>住所:<\/b>\s*(.*)\s*<\/div>/m =~ content
        @place["description"]  = $1.chomp    
      end
    end
  end

  def text(text)
    case @context.last
    when "coordinates"
      if @context == @@coordinate_context then
        /([-0-9\.]+),([-0-9\.]+)(?:,([-0-9\.]+))/m =~ text
        @place["lng"] = $1.to_f
        @place["lat"] = $2.to_f
      end
    end
  end
end

listener = KmlListener.new
Document.parse_stream($stdin, listener)

db = SQLite3::Database.new("shelter.db")
db.execute("create table if not exists places (id integer primary key, name text, lat real, lng real, detail text);")
db.execute("create table if not exists geohash (geohash text primary key, lat real, lng real, glat real, glng real, length integer, center_count integer, neighbor_count integer);")
db.execute("create table if not exists place_geohashes (places_id integer references places(id), geohash text references geohash(geohash), center integer);")
db.execute("delete from place_geohashes;")
db.execute("delete from geohash;")
db.execute("delete from places;")
db.execute("vacuum;")
db.transaction do
  array = listener.array
  array.each {|place|
    next if !place.key?('lat') || !place.key?('lng')
    p place["id"]
    db.execute("insert or replace into places values (:id, :name, :lat, :lng, :detail);", place["id"], place["name"], place["lat"], place["lng"], place["description"])
    [4, 5, 6].each {|l|
      center = GeoHash.encode(place["lat"], place["lng"], l)
      center_pos = GeoHash.decode(center)
      db.execute("insert or replace into geohash values (:geohash, :lat, :lng, 0, 0, :length, 0, 0)", center, (center_pos[0][0]+center_pos[1][0])/2, (center_pos[0][1]+center_pos[1][1])/2, l)
      GeoHash.neighbors(center).each {|n|
        n_pos = GeoHash.decode(n)
        db.execute("insert or replace into geohash values (:geohash, :lat, :lng, 0, 0, :length, 0, 0)", n, (n_pos[0][0]+n_pos[1][0])/2, (n_pos[0][1]+n_pos[1][1])/2, l)
      }
    }
  }
  array.each {|place|
    next if !place.key?('lat') || !place.key?('lng')
    p place["id"]
    [4, 5, 6].each {|l|
      center = GeoHash.encode(place["lat"], place["lng"], l)
      db.execute("insert into place_geohashes values (:placesid, :geohash, 1)", place["id"], center)
      db.execute("update geohash set center_count=center_count+1, glat=(glat*center_count+:lat)/(center_count+1), glng=(glng*center_count+:lng)/(center_count+1) where geohash=:geohash", place["lat"], place["lng"], center)
      GeoHash.neighbors(center).each {|n|
        n_pos = GeoHash.decode(n)
        db.execute("insert into place_geohashes values (:placesid, :geohash, 0)", place["id"], n)
        db.execute("update geohash set neighbor_count=neighbor_count+1 where geohash=:geohash", n)
      }
    }
  }
end
db.execute("vacuum;")
db.close