BusStopMashup.py

   1 #
   2 # vi: set softtabstop=4 shiftwidth=4 tabstop=8 expandtab:
   3
   4 """Downloads Google Maps bus stop locations using the official mashup."""
   5
   6 import urllib
   7 import urllib2
   8 import sys
   9 import re
  10
  11 class StopRoute(object):
  12     def __init__(self, number, direction):
  13         self.number = number        # route number
  14         self.direction = direction  # route direction, seems to be 10 or 11
  15
  16     def __str__(self):
  17         return "route(%s,%s)" % (self.number, self.direction)
  18
  19 class StopLocation(object):
  20     def __init__(self, latitude, longitude):
  21         self.latitude = float(latitude)
  22         self.longitude = float(longitude)
  23
  24     def __str__(self):
  25         return "loc (%s,%s)" % (self.latitude, self.longitude)
  26
  27 class BaseStop(object):
  28     def __init__(self):
  29         self.number = None          # 560-1000 number
  30                                     # This is not necessarily unique - there
  31                                     # are multiple stops #3034, for each
  32                                     # platform at Billings Bridge
  33         self.code = None            # e.g. WD360 - looks to be unique
  34         self.name = None            # usually an intersection
  35         self.location = None
  36
  37     def basestr(self):
  38         return "560 %s,%s,%s,%s" % (self.number, self.code,
  39                                     self.name, self.location)
  40
  41     def __str__(self):
  42         return "basestop (%s)" % self.basestr()
  43
  44 class Stop(BaseStop):
  45     """A regular bus stop."""
  46     def __init__(self):
  47         super(Stop, self).__init__()
  48         self.routes = []            # list of StopRoutes at this stop
  49
  50     def __str__(self):
  51         return "stop (%s): %s" % (
  52                     self.basestr(), ", ".join([str(r) for r in self.routes]))
  53
  54 class Station(BaseStop):
  55     """A Transitway or O-Train station."""
  56     def __init__(self):
  57         super(Station, self).__init__()
  58         # code will be short (3- or 4-letter) station identifier
  59         # name will be station name
  60
  61     def __str__(self):
  62         return "station (%s)" % self.basestr()
  63
  64 class UnknownStopType(BaseStop):
  65     def __init__(self, location, html):
  66         self.location = location
  67         self.html = html
  68
  69     def __str__(self):
  70         return "UNKNOWN (%s, %s)" % (self.location, self.html)
  71
  72 class HomeLocation(BaseStop):
  73     def __init__(self, location, requestedAddress, respondedAddress):
  74         self.location = location
  75         self.requestedAddress = requestedAddress
  76         self.respondedAddress = respondedAddress
  77
  78     def __str__(self):
  79         return "HOME: (asked %s, got %s, %s)" % (
  80                 self.requestedAddress, self.respondedAddress, self.location)
  81
  82 class Marker(object):
  83     def __init__(self, location, html):
  84         self.location = location
  85         self.html = html
  86
  87     def __str__(self):
  88         return "marker (%s, %s)" % (self.location, self.html)
  89
  90 class Client(object):
  91     def findStops(self, address):
  92         for marker in self._findMarkers(self._grabHTML(address)):
  93             for (rx, func) in (
  94                     (_station_rx, self._parseStation),
  95                     (_stop_rx, self._parseStop),
  96                     (_home_rx, self._parseHome)):
  97
  98                 match = rx.search(marker.html)
  99                 if match is not None:
 100                     ret = func(address, marker, match)
 101                     if ret is not None:
 102                         yield ret
 103                     break
 104             if match is None:
 105                 yield UnknownStopType(marker.location, marker.html)
 106
 107     def _grabHTML(self, address):
 108         params = { 'address': address }
 109         params.update(FIXED_PARAMS)
 110         f = urllib2.urlopen(URL, urllib.urlencode(params))
 111
 112         html = ""
 113         for line in f:
 114             html += line
 115         f.close()
 116
 117         return html
 118
 119     def _findMarkers(self, html):
 120         for m in _marker_rx.finditer(html):
 121             yield Marker(StopLocation(m.group("latitude"),
 122                                       m.group("longitude")),
 123                          m.group("html"))
 124
 125     def _fillBaseStop(self, marker, match, stop):
 126         stop.number = match.group("stopnum")
 127         stop.code = match.group("code")
 128         stop.name = match.group("name")
 129         stop.location = marker.location
 130
 131     def _parseStation(self, address, marker, match):
 132         stop = Station()
 133         self._fillBaseStop(marker, match, stop)
 134         return stop
 135
 136     def _parseStop(self, address, marker, match):
 137         stop = Stop()
 138         self._fillBaseStop(marker, match, stop)
 139
 140         for m in _stop_route_rx.finditer(marker.html):
 141             stop.routes.append(
 142                     StopRoute(m.group("routenum"), m.group("direction")))
 143         return stop
 144
 145     def _parseHome(self, address, marker, match):
 146         return HomeLocation(marker.location, address, match.group("homeaddr"))
 147
 148 URL = "http://www.octranspo.com/maps/busstops/imap.asp"
 149 FIXED_PARAMS = { 'page': 'search' }
 150
 151 # Any type of marker
 152 # var marker = createMarker(new GPoint(..., ...), "blah blah");
 153 # note: gmaps api v1 uses GPoint(long, lat) instead of GLatLng(lat, long)
 154 _marker_re = (r'createMarker\(new GPoint\('
 155               r'(?P<longitude>[^,]+),\s*(?P<latitude>[^,]+)\),\s*'
 156               # ?s: DOTALL: . matches \n
 157               # *? is non-greedy
 158               r'"(?s)(?P<html>.*?)"\);\r?\n')
 159 _marker_rx = re.compile(_marker_re)
 160
 161
 162 # Marker for a transitway station
 163 # <span><strong><b>613-560-1000 plus <a href='iframe.asp?route=busstop&INFO_PHONE=3034' target='iframe'>3034</a><br><a href='iframe.asp?route=bus_station&INFO_PHONE=3034&station_name=BILLINGS BRIDGE&station_id=BIB' target='iframe'>BILLINGS BRIDGE</b></strong></span>
 164 # (?i): Make it case-insensitive
 165 # For station name, assume they'll URL-encode it properly someday
 166 _station_re = (r'(?i)INFO_PHONE=(?P<stopnum>\w+)'
 167                r'.*station_name=(?P<name>[\w\s%+]+)'
 168                r'.*station_id=(?P<code>\w+)')
 169 _station_rx = re.compile(_station_re)
 170
 171 # Marker for a regular stop
 172 # <span><strong><b>613-560-1000 plus <a href='iframe.asp?route=busstop&INFO_PHONE=4897' target='iframe'>4897</a></b></strong><small>  (RA040)</small><br>BANK / TRANSITWAY<br><a href='iframe.asp?route=1&dir=10' target='iframe'>1</a> <a href='iframe.asp?route=5&dir=10' target='iframe'>5</a> <a href='iframe.asp?route=111&dir=10' target='iframe'>111</a> <a href='iframe.asp?route=141&dir=11' target='iframe'>141</a> <a href='iframe.asp?route=148&dir=10' target='iframe'>148</a> </span>
 173 _stop_re = (r'(?i)INFO_PHONE=(?P<stopnum>\w+)'
 174             r'.*\((?P<code>[A-Z]{2,3}[0-9]{3})\)'
 175             # The stop name could be BLAH, BLAH / BLAH, BLAH STOP/ARR?T 1A.
 176             # Look for some English letters and numbers at the start, then let
 177             # anything in.
 178             r'.*\<br\>(?P<name>\w[^<>]+)\<br\>')
 179 _stop_rx = re.compile(_stop_re)
 180
 181 _stop_route_re = (r'(?i)route=(?P<routenum>\w+)&dir=(?P<direction>\w+)')
 182 _stop_route_rx = re.compile(_stop_route_re)
 183
 184 # Marker for the current location
 185 # <span><strong>HERON RD & DATA CENTRE RD, OTTAWA, ON, CANADA</strong></span>"
 186 _home_re = (r'(?i)(?P<homeaddr>[^<>]+), CANADA')
 187 _home_rx = re.compile(_home_re)
 188
 189 def main(argv=None):
 190     if argv is None:
 191         argv = sys.argv
 192
 193     cmdstr = " ".join(argv[1:])
 194     for s in Client().findStops(cmdstr):
 195         print s
 196
 197 if __name__ == '__main__':
 198     sys.exit(main())