Planner.py

   1 #
   2 # vi: set softtabstop=4 shiftwidth=4 tabstop=8 expandtab:
   3
   4 """A frontend for the OC Transpo Travel Planner."""
   5
   6 import cookielib
   7 import urllib2
   8 import urllib
   9 import sre
  10
  11 import Itinerary
  12
  13 def plan(start, end, time):
  14     """Plans a route between two Locations at a certain PlanTime."""
  15     planner = TravelPlannerClient()
  16     planner.feedStartLocation(start)
  17     planner.feedEndLocation(end)
  18     html = planner.feedTime(time)
  19     return Itinerary.Itinerary(start, end, time, html)
  20
  21 class TravelPlannerException(Exception):
  22     """Generic parent exception for anything thrown by us, as opposed to
  23        things thrown by the network layer."""
  24
  25     def __init__(self, value):
  26         Exception.__init__(self, value)
  27
  28 class InvalidLocationException(TravelPlannerException):
  29     """Thrown when a source or destination can't be found.
  30
  31        The "args" value of the exception is the original Location."""
  32
  33     def _init(self, location):
  34         TravelPlannerException.__init__(self, location)
  35
  36 class TravelPlannerClient:
  37     def __init__(self):
  38         # Set up a cookie-aware client.
  39         self.cj = cookielib.CookieJar()
  40         self.opener = urllib2.build_opener(urllib2.HTTPCookieProcessor(self.cj))
  41
  42         # Initialize the session.
  43         r = self._sendRequest(self.START_PAGE, None)
  44
  45     def feedStartLocation(self, loc):
  46         # name="tp_fromAddress" method="get" action="FromAddress.oci"
  47         # fromAddress text
  48         # streetType=Other
  49         params = (("fromAddress", loc), ("streetType", "Other"))
  50         r = self._sendRequest("FromAddress.oci", params)
  51
  52     def feedEndLocation(self, loc):
  53         params = (("toAddress", loc), ("streetType", "Other"))
  54         r = self._sendRequest("ToAddress.oci", params)
  55
  56     def feedTime(self, time):
  57         # name="tp_time" action="SelectTime.oci"
  58         r = self._sendRequest("SelectTime.oci", time.toPlannerParams())
  59
  60         return r
  61
  62     def _sendRequest(self, page, params):
  63         url = self.URL_BASE + page
  64         if params is not None:
  65             url += "?" + urllib.urlencode(params)
  66
  67         response = self.opener.open(url)
  68         self._checkObviousBadness(response)
  69
  70         html = self._grabLimitedResponse(response)
  71         self._scanForError(html)
  72         return html
  73
  74     def _checkObviousBadness(self, response):
  75         if response.code != 200:
  76             raise TravelPlannerException("Got HTTP " + response.code
  77                                          + " error from server")
  78         if "errorPage.oci" in response.geturl():
  79             raise TravelPlannerException("Redirected to error page")
  80
  81     def _grabLimitedResponse(self, response):
  82         count = 0
  83
  84         accum = ""
  85         for line in response:
  86             count += len(line)
  87             if (count <= self.RESPONSE_SIZE_LIMIT):
  88                 accum += line
  89             else:
  90                 break
  91         return accum
  92
  93
  94
  95     # Regular expressions: probably the worst way to parse HTML
  96     # Probably the best thing to put in your breakfast cereal.
  97
  98     def _scanForError(self, text):
  99         match = _error_rx.search(text)
 100         if match:
 101             raise TravelPlannerException(match.group("msg"))
 102
 103     URL_BASE = "http://www.octranspo.com/tps/jnot/"
 104     START_PAGE = "startEN.oci"
 105     RESPONSE_SIZE_LIMIT = 100000
 106
 107
 108
 109 # Scan for an error.
 110 # <table cellpadding="0" cellspacing="0" summary="Warning message" class="warning" width="85%">
 111 #   <tr>
 112 #       <td><img src="tripPlanning/images/imgWarning.gif"></td>
 113 #       <td>The address you specified was not found.  Please enter another.</td>
 114 #   </tr>
 115 # </table>
 116 _error_re = ('<table[^>]*class="(?:warning|error)[^>]*>\s*'
 117                 '<tr>(?:\s*<td>\s*<img[^>]*>\s*</td>)?'
 118                     # ?s: DOTALL: . matches \n
 119                     # *? is non-greedy
 120                     '\s*<td>(?s)\s*(?P<msg>[\d\D]*?)\s*</td>')
 121 _error_rx = sre.compile(_error_re)