vcs.maemo.org Git - pywienerlinien/blob - gotovienna/realtime.py

   1 # -*- coding: utf-8 -*-
   2
   3 from BeautifulSoup import BeautifulSoup
   4 from urllib2 import urlopen
   5 from datetime import time
   6 import argparse
   7 import re
   8
   9 from gotovienna import defaults
  10
  11 class ITipParser:
  12     def __init__(self):
  13         self._stations = {}
  14         self._lines = {}
  15
  16     def get_stations(self, name):
  17         """ Get station by direction
  18         {'Directionname': [('Station name', 'url')]}
  19         """
  20         if not self._stations.has_key(name):
  21             st = {}
  22
  23             if not self.lines.has_key(name):
  24                 return None
  25
  26             bs = BeautifulSoup(urlopen(self.lines[name]))
  27             tables = bs.findAll('table', {'class': 'text_10pix'})
  28             for i in range(2):
  29                 dir = tables[i].div.contents[-1].strip('&nbsp;')
  30
  31                 sta = []
  32                 for tr in tables[i].findAll('tr', {'onmouseout': 'obj_unhighlight(this);'}):
  33                     if tr.a:
  34                         sta.append((tr.a.text, defaults.line_overview + tr.a['href']))
  35                     else:
  36                         sta.append((tr.text.strip('&nbsp;'), None))
  37
  38                 st[dir] = sta
  39             self._stations[name] = st
  40
  41         return self._stations[name]
  42
  43     @property
  44     def lines(self):
  45         """ Dictionary of Line names with url as value
  46         """
  47         if not self._lines:
  48             bs = BeautifulSoup(urlopen(defaults.line_overview))
  49             # get tables
  50             lines = bs.findAll('td', {'class': 'linie'})
  51
  52             for line in lines:
  53                 if line.a:
  54                     href = defaults.line_overview + line.a['href']
  55                     if line.text:
  56                         self._lines[line.text] = href
  57                     elif line.img:
  58                         self._lines[line.img['alt']] = href
  59
  60         return self._lines
  61
  62     def get_departures(self, url):
  63         """ Get list of next departures
  64         integer if time until next departure
  65         time if time of next departure
  66         """
  67
  68         #TODO parse line name and direction for station site parsing
  69
  70         if not url:
  71             # FIXME prevent from calling this method with None
  72             return []
  73
  74         bs = BeautifulSoup(urlopen(url))
  75         result_lines = bs.findAll('table')[-1].findAll('tr')
  76
  77         dep = []
  78         for tr in result_lines[1:]:
  79             th = tr.findAll('th')
  80             if len(th) < 2:
  81                 #TODO replace with logger
  82                 print "[DEBUG] Unable to find th in:\n%s" % str(tr)
  83                 continue
  84
  85             # parse time
  86             time = th[-2].text.split(' ')
  87             if len(time) < 2:
  88                 print 'Invalid time: %s' % time
  89                 continue
  90
  91             time = time[1]
  92
  93             if time.find('rze...') >= 0:
  94                     dep.append(0)
  95             elif time.isdigit():
  96                 # if time to next departure in cell convert to int
  97                 dep.append(int(time))
  98             else:
  99                 # check if time of next departue in cell
 100                 t = time.strip('&nbsp;').split(':')
 101                 if len(t) == 2 and all(map(lambda x: x.isdigit(), t)):
 102                     t = map(int, t)
 103                     dep.append(time(*t))
 104                 else:
 105                     # Unexpected content
 106                     #TODO replace with logger
 107                     print "[DEBUG] Invalid data:\n%s" % time
 108
 109         return dep
 110