vcs.maemo.org Git - pywienerlinien/blob - iTip.py

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