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 __repr__(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 __repr__(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         self.requestedAddress = None  # search address that returned this stop
  38
  39     def basestr(self):
  40         return "560 %s,%s,%s,%s" % (
  41                 self.number, repr(self.code), repr(self.name), self.location)
  42
  43     def __repr__(self):
  44         return "basestop (%s)" % self.basestr()
  45
  46 class Stop(BaseStop):
  47     """A regular bus stop."""
  48     def __init__(self):
  49         super(Stop, self).__init__()
  50         self.routes = []            # list of StopRoutes at this stop
  51
  52     def __repr__(self):
  53         return "stop (%s): %s" % (
  54                     self.basestr(), ", ".join([str(r) for r in self.routes]))
  55
  56 class Station(BaseStop):
  57     """A Transitway or O-Train station."""
  58     def __init__(self):
  59         super(Station, self).__init__()
  60         # code will be short (3- or 4-letter) station identifier
  61         # name will be station name
  62
  63     def __repr__(self):
  64         return "station (%s)" % self.basestr()
  65
  66 class UnknownStopType(BaseStop):
  67     def __init__(self, location, html):
  68         super(UnknownStopType, self).__init__()
  69         self.location = location
  70         self.html = html
  71
  72     def __repr__(self):
  73         return "UNKNOWN (%s, %s)" % (self.location, repr(self.html))
  74
  75 class HomeLocation(BaseStop):
  76     def __init__(self, location, requestedAddress, respondedAddress):
  77         super(HomeLocation, self).__init__()
  78         self.location = location
  79         self.requestedAddress = requestedAddress
  80         self.respondedAddress = respondedAddress
  81
  82     def __repr__(self):
  83         return "HOME: (asked %s, got %s, %s)" % (
  84                 repr(self.requestedAddress), repr(self.respondedAddress),
  85                 self.location)
  86
  87 class Marker(object):
  88     def __init__(self, location, html):
  89         self.location = location
  90         self.html = html
  91
  92     def __repr__(self):
  93         return "marker (%s, %s)" % (self.location, repr(self.html))
  94
  95 class InvalidAddressException(Exception):
  96     def __init__(self, address):
  97         super(InvalidAddressException, self).__init__(address)
  98
  99 class Client(object):
 100     def findStops(self, address):
 101         html = self._grabHTML(address)
 102         self._checkForErrors(html, address)
 103         for marker in self._findMarkers(html):
 104             for (rx, func) in (
 105                     (_station_rx, self._parseStation),
 106                     (_stop_rx, self._parseStop),
 107                     (_home_rx, self._parseHome)):
 108
 109                 match = rx.search(marker.html)
 110                 if match is not None:
 111                     ret = func(address, marker, match)
 112                     if ret is not None:
 113                         yield ret
 114                     break
 115             if match is None:
 116                 yield UnknownStopType(marker.location, marker.html)
 117
 118     def _grabHTML(self, address):
 119         params = { 'address': address }
 120         params.update(FIXED_PARAMS)
 121         f = urllib2.urlopen(URL, urllib.urlencode(params))
 122
 123         html = ""
 124         for line in f:
 125             html += line
 126         f.close()
 127
 128         return html
 129
 130     def _checkForErrors(self, html, address):
 131         if _error_rx.search(html) is not None:
 132             raise InvalidAddressException(address)
 133
 134     def _findMarkers(self, html):
 135         for m in _marker_rx.finditer(html):
 136             yield Marker(StopLocation(m.group("latitude"),
 137                                       m.group("longitude")),
 138                          m.group("html").decode(REAL_ENCODING))
 139
 140     def _fillBaseStop(self, address, marker, match, stop):
 141         stop.number = match.group("stopnum")
 142         stop.code = match.group("code")
 143         stop.name = match.group("name").strip()
 144         stop.location = marker.location
 145         stop.requestedAddress = address
 146
 147     def _parseStation(self, address, marker, match):
 148         stop = Station()
 149         self._fillBaseStop(address, marker, match, stop)
 150         return stop
 151
 152     def _parseStop(self, address, marker, match):
 153         stop = Stop()
 154         self._fillBaseStop(address, marker, match, stop)
 155
 156         for m in _stop_route_rx.finditer(marker.html):
 157             stop.routes.append(
 158                     StopRoute(m.group("routenum"), m.group("direction")))
 159         return stop
 160
 161     def _parseHome(self, address, marker, match):
 162         return HomeLocation(marker.location, address, match.group("homeaddr"))
 163
 164 URL = "http://www.octranspo.com/maps/busstops/imap.asp"
 165 FIXED_PARAMS = { 'page': 'search' }
 166
 167 # There's a meta tag that says it's UTF-8, but it turns out it's latin-1.
 168 REAL_ENCODING = "latin-1"
 169
 170 # Any type of marker
 171 # var marker = createMarker(new GPoint(..., ...), "blah blah");
 172 # note: gmaps api v1 uses GPoint(long, lat) instead of GLatLng(lat, long)
 173 _marker_re = (r'createMarker\(new GPoint\('
 174               r'(?P<longitude>[^,]+),\s*(?P<latitude>[^,]+)\),\s*'
 175               # ?s: DOTALL: . matches \n
 176               # *? is non-greedy
 177               r'"(?s)(?P<html>.*?)"\);\r?\n')
 178 _marker_rx = re.compile(_marker_re)
 179
 180
 181 # Marker for a transitway station
 182 # <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>
 183 # (?i): Make it case-insensitive
 184 # For station name, assume they'll URL-encode it properly someday
 185 _station_re = (r'(?i)INFO_PHONE=(?P<stopnum>\w+)'
 186                r'.*station_name=(?P<name>[^<>&]+)'
 187                r'.*station_id=(?P<code>\w+)')
 188 _station_rx = re.compile(_station_re)
 189
 190 # Marker for a regular stop
 191 # <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>
 192 _stop_re = (r'(?i)INFO_PHONE=(?P<stopnum>\w*)'
 193             r'.*?\<small\>\s*\((?P<code>[^)]+)\)'
 194             r'.*\<br\>(?P<name>[^<>]+)\<br\>')
 195 _stop_rx = re.compile(_stop_re)
 196
 197 _stop_route_re = (r'(?i)route=(?P<routenum>\w+)&dir=(?P<direction>\w+)')
 198 _stop_route_rx = re.compile(_stop_route_re)
 199
 200 # Marker for the current location
 201 # <span><strong>HERON RD & DATA CENTRE RD, OTTAWA, ON, CANADA</strong></span>"
 202 _home_re = (r'(?i)(?P<homeaddr>[^<>]+), CANADA')
 203 _home_rx = re.compile(_home_re)
 204
 205 _error_re = (r'G_GEO_UNKNOWN_ADDRESS')
 206 _error_rx = re.compile(_error_re)
 207
 208 def main(argv=None):
 209     if argv is None:
 210         argv = sys.argv
 211
 212     cmdstr = " ".join(argv[1:])
 213     for s in Client().findStops(cmdstr):
 214         print s
 215
 216 if __name__ == '__main__':
 217     sys.exit(main())