Python!

DVB-Scraping.mw

@@ -1,9 +1,11 @@
 [[Kategorie:Projekt]]
 
+=Ruby=
+
 {{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=
+==Telnet-Haltestellenmonitor==
 <source lang="ruby">#!/usr/bin/env ruby
 require 'net/http'
 require 'socket'
@@ -131,7 +133,7 @@ end
 Und dann:
  telnet localhost 65023
 
-=NCurses Monitor=
+==NCurses Monitor==
 Der VVO stellt jetzt ja für seine Widgets eine JSON(?)-Variante der Daten zur Verfügung.
 
 <source lang="ruby">
@@ -245,6 +247,176 @@ if __FILE__ == $0
 end
 </source>
 
+=Python=
+
+==CLI-Interface==
+<source lang="python">
+#!/usr/bin/python
+
+import sys
+
+from urllib import urlencode, urlopen
+from optparse import OptionParser
+
+from BeautifulSoup import BeautifulStoneSoup
+import simplejson
+
+widgets_base_url = "http://widgets.vvo-online.de/abfahrtsmonitor/"
+
+def get_connections(stop=None, town=None, time=None):
+    """
+    Get the next connections at *stop* in *town* *time* minutes from now.
+    """
+    query_params = []
+
+    if stop is not None:
+        query_params.append(("hst", stop))
+    if town is not None:
+        query_params.append(("ort", town))
+    if time is not None:
+        query_params.append(("vz", time))
+
+    query_url = widgets_base_url + "Abfahrten.do?" + urlencode(query_params)
+
+    page_data = urlopen(query_url).read()
+    connections_soup = BeautifulStoneSoup(page_data, convertEntities="html")
+    connections_data = connections_soup.contents[0]
+
+    connections = simplejson.loads(connections_data)
+
+    return connections
+
+
+def find_stops(stop, town=None):
+    """
+    Get stops with the given name in *town*.
+    """
+    query_params = [("hst", stop)]
+
+    if town is not None:
+        query_params.append(("ort", town))
+
+    query_url = widgets_base_url + "Haltestelle.do?" + urlencode(query_params)
+
+    page_data = urlopen(query_url).read()
+    stops_soup = BeautifulStoneSoup(page_data, convertEntities="html")
+    stops_data = stops_soup.contents[0]
+
+    towns, stops = simplejson.loads(stops_data)
+
+    return towns, stops
+
+
+def format_connections(connections):
+    """
+    Format a list of connections into a nice table.  Returns a generator for
+    table's rows.
+    """
+
+    destination_column_length = max(23, *(len(d) for _, d, _ in connections))
+    line_name_column_length = max(5, *(len(l) for l, _, _ in connections))
+    line_format = "%-" + str(line_name_column_length) + "s | %" \
+            + str(destination_column_length) + "s | %7s"
+    header_line = line_format % ("line", "destination", "arrival")
+
+    yield header_line
+    yield "-" * line_name_column_length + "-+-" \
+            + "-" * destination_column_length + "-+-" \
+            + "-" * 7
+
+    for line_name, destination, time in connections:
+        yield line_format % (line_name, destination, time)
+
+
+def print_connections(stop, town, connections, limit=None):
+    """
+    Print *connections* at *stop* in *town* to stdout.  If a limit is given
+    only that many connections are printed otherwise all.
+    """
+
+    if len(connections) == 0:
+        print "No connections at %s in %s." % (stop, town)
+        sys.exit(1)
+
+    if town is not None:
+        print "Next connections at %s in %s:" % (stop, town)
+    else:
+        print "Next connections at %s:" % (stop,)
+
+    print
+
+    if limit:
+        connections_table = format_connections(connections[:limit])
+    else:
+        connections_table = format_connections(connections)
+
+    for line in connections_table:
+        print line
+
+
+def main():
+    """
+    Main function.
+    """
+
+    option_parser = OptionParser(
+            usage="%prog [options] [<town>] <stop>")
+    option_parser.add_option("-l", "--limit",
+            help="maximum number of connections to display",
+            type="int",
+            default=5)
+    option_parser.add_option("-t", "--time",
+            help="minimum time to departure",
+            type="int",
+            default=None)
+    option_parser.add_option("-k", "--no-lookup",
+            help="do not look up stop name",
+            action="store_false",
+            dest="lookup_stop",
+            default=True)
+
+    options, args = option_parser.parse_args()
+
+    # sanitize options
+    if options.limit < 0:
+        options.limit = None
+    if options.time < 0:
+        options.time = None
+
+    if len(args) == 1:
+        stop = args[0]
+        town = None
+    elif len(args) == 2:
+        town, stop = args
+    else:
+        option_parser.error("Not enough arguments")
+
+    if options.lookup_stop:
+        towns, stops = find_stops(stop, town)
+
+        if len(towns) == 0:
+            print "No town named '%s'." % (town,)
+            sys.exit(1)
+
+        if len(stops) == 0:
+            print "No stop named '%s' in the following towns:" % (stop,)
+            print "\n".join("    "+t[0] for t in towns)
+            sys.exit(1)
+
+        for stop_name, town, stop_id in stops:
+            connections = get_connections(stop_id, time=options.time)
+
+            print_connections(stop_name, town, connections, options.limit)
+            print
+    else:
+        connections = get_connections(stop, town, options.time)
+        print_connections(stop, town, connections, options.limit)
+
+
+if __name__ == "__main__":
+    main()
+</source>
+
 [[Kategorie:Ruby]]
 
 {{Rübÿ Spëëd Mëtäl Cödïng}}