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