c3d2-wiki/DVB-Scraping.mw

[[Kategorie:Projekt]]

{{Broken|
Reason=Die Struktur der WAP-Seiten war nicht permanent und wurde in der Zwischenzeit geändert. Jemand muss auf die neuen URLs anpassen.|}}

=Telnet-Haltestellenmonitor=
<source lang="ruby">#!/usr/bin/env ruby
require 'net/http'
require 'socket'
require 'rexml/document'
class MultipleStations < RuntimeError
  def initialize(stations)
    @stations = stations
  end
  def to_s
    "Mehrere mögliche Haltestellen: " + @stations.join(', ')
  end
end
class StationResult
  def initialize(card)
    strong_n = 0
    card.each_element('p/strong') { |e|
      case strong_n
        when 0
          @name = e.text
        when 1
          @time = e.text
      end
      strong_n += 1
    }
    @trams = []
    card.to_s.scan(/br\/>(\d+:\d+\*?) (.+?)<br\/>-> (.+?)</) { |time,tram,direction|
      @trams << [time, tram, direction]
    }
  end
  def to_s
    column_widths = [4, 5, 4]
    @trams.each { |a|
      a.each_with_index { |b,i|
        column_widths[i] = b.size if b.size > column_widths[i]
      }
    }
    "\n\n#{@name}, #{@time}:\n\n" +
    'Zeit'.ljust(column_widths[0]) + ' | ' +
      'Linie'.ljust(column_widths[1]) + ' | ' +
      'Ziel'.ljust(column_widths[2]) + "\n" +
    ('-' * column_widths[0]) + '-+-' +
      ('-' * column_widths[1]) + '-+-' +
      ('-' * column_widths[2]) + "\n" +
    @trams.collect { |time,tram,direction|
      time.ljust(column_widths[0]) + ' | ' +
        tram.ljust(column_widths[1]) + ' | ' +
        direction.ljust(column_widths[2])
    }.join("\n")
  end
end
class ClientHandler
  def initialize(socket)
    @socket = socket
    puts "#{address} connected"
    Thread.new {
      begin
        handle
      rescue Exception => e
        @socket.puts("Fehler: #{e}")
      ensure
        @socket.close
      end
    }
  end
  def address
    if @socket.peeraddr[0] == 'AF_INET6'
      "[#{@socket.peeraddr[3]}]"
    else
      "#{@socket.peeraddr[3]}"
    end +
    ":#{@socket.peeraddr[1]}"
  end
  def ask_haltestellenmonitor(station)
    param = { :station => station,
              :action => :check,
              :time => Time.new.strftime('%H:%M'),
              :date => Time.new.strftime('%d.%m.%Y')
            }
    param_s = param.collect { |k,v| "#{k}=#{v}" }.join('&')
    param_s.gsub!(/ /, '+')
    res = Net::HTTP.start('wap.dvbag.de') { |http|
      http.get('/wapVVO/wap-rbl.php?' + param_s)
    }
    if res.kind_of? Net::HTTPSuccess
      wml = REXML::Document.new(res.body).root
      card = nil
      wml.each_element('/wml/card') { |c| card = c }
      if card
        if card.attributes['id'] == 'liste'
          stations = []
          card.each_element('p/select/option') { |option|
            stations << option.text
          }
          raise MultipleStations.new(stations)
        elsif card.attributes['id'] == 'result'
          StationResult.new(card).to_s
        else
          raise "Unexpected card/@id: #{card.attributes['id']}"
        end 
      else
        raise "No card found in result document"
      end
    else
      raise "#{res.class}"
    end
  end
  def handle
    @socket.print "Hallo #{address}\n\nHaltestelle: "
    @socket.flush
    haltestelle = @socket.gets
    if haltestelle
      haltestelle.strip!
      puts "#{address} asks for #{haltestelle.inspect}"
      @socket.puts "Anfrage nach #{haltestelle}..."
      @socket.puts ask_haltestellenmonitor(haltestelle)
    end
  end
end
serv = TCPServer.new('0.0.0.0', 65023)
while client = serv.accept
  ClientHandler.new(client)
end
</source>

Und dann:
 telnet localhost 65023

=NCurses Monitor=
Der VVO stellt jetzt ja für seine Widgets eine JSON(?)-Variante der Daten zur Verfügung.

<source lang="ruby">

require 'net/http'
require 'ncurses'

class DvbAbfahrt
  def initialize
    @BASEURI = "http://widgets.vvo-online.de/abfahrtsmonitor/Abfahrten.do?ort=ORT&hst=HST&vz=VZ"
        
  end
  
  def fetch(ort, hst, vz=0)
		# TODO exceptionhandling: timeout
    vz = vz.to_s
		ort = URI.escape ort
		hst = URI.escape hst
    uri = @BASEURI.gsub(/ORT/, ort).gsub(/HST/,hst).gsub(/VZ/, vz)
    res = Net::HTTP.get(URI.parse(uri))
    res = umlauts_hack res 
   	arr = res.scan(/(\d+),([^,]{1,}),(\d+)/) 
  end

	def umlauts_hack(s)
		repl = [
			["&quot;", ''],
			["&#252;", "ü"],
			["&#246;", "ö"],
			["&#223;", "ß"]
		]
		repl.each do |r|
			s.gsub!(r[0], r[1])
		end
		return s
	end
  
end

class Monitor
	
	def initialize(ort, hst, vz=0) 
		Ncurses::initscr
		Ncurses::start_color
		Ncurses::init_pair(1, Ncurses::COLOR_YELLOW, Ncurses::COLOR_BLACK);
		Ncurses::attron(Ncurses::COLOR_PAIR(1));
		Ncurses::curs_set(0)
		Ncurses::move(0,0)
		Ncurses::printw "Loading..."
		Ncurses::refresh
		
		@sep = "|"
		@timeout = 30	
		
		@wlno = 3
		@wlname = 20 
		@weta = 3
		
		@ort = ort
		@hst = hst
		@vz = vz
		@dvb = DvbAbfahrt.new
		@lines = 0
	end

	def print_title
		Ncurses::attron(Ncurses::A_REVERSE | Ncurses::A_BOLD);
		Ncurses::mvprintw(0,0, @hst.center(@wlno+@wlname+@weta+2)) # +2 seperators
		Ncurses::attroff(Ncurses::A_REVERSE | Ncurses::A_BOLD);
	end

	def mainloop
		print_title
		while true
			info = @dvb.fetch(@ort, @hst, @vz)
			
		# clearing old lines if needed
			
			if @lines > info.size
				Ncurses::clear
				print_title
			end

			@lines = info.size

			info.each_index do |i|
				
				Ncurses::mvprintw(i+1, 0, info[i][0].rjust(@wlno) + @sep + info[i][1][0..@wlname].center(@wlname) + @sep + info[i][2].rjust(@weta))
			end
			Ncurses::refresh
			Ncurses::move(0,0)
			sleep @timeout
		end
	end
end



if __FILE__ == $0
	begin
		if !ARGV.empty?
			m = Monitor.new("Dresden", ARGV[0])
			m.mainloop
		else
			puts "USAGE #{$0} <HALTESTELLENNAME>"
		end
	ensure
		Ncurses::attroff(Ncurses::COLOR_PAIR(1));
		Ncurses::endwin
	end
end
</source>

[[Kategorie:Ruby]]

{{Rübÿ Spëëd Mëtäl Cödïng}}