Generate iCal Feed from AHA Hannover website

_etc_lighttpd_sites_cal.conf

       $HTTP["host"] == "cal.wgchaos.info" {
            server.name          = "cal.wgchaos.info"
            var.dir              = "info/wgchaos/cal"
            server.document-root = "/var/www/info/wgchaos/cal/html"
            accesslog.filename   = "/var/log/lighttpd/access-cal.wgchaos.info.log"


            $HTTP["host"] != "cal.wgchaos.info" {
                $HTTP["host"] =~ "^(.*)$" {
                    url.redirect = (
                        "^/(.*)" => "http://cal.wgchaos.info/$1",
                    )
                } # end of $HTTP["host"] =~ "^(.*)$"
            } # end of $HTTP["host"] != "cal.wgchaos.info"

            $HTTP["url"] == "/aha" {
                cgi.assign = (
                    "" => "",
                )
            } # end of $HTTP["url"] == "/aha"
        } # end of $HTTP["host"] == "cal.wgchaos.info"

generate.py

#!/usr/bin/env python

from datetime import datetime, timedelta
import locale
import time
import cgi

from lxml.html.soupparser import fromstring
import requests
import vobject


# german locale for dateparsing
locale.setlocale(locale.LC_TIME, ('de_DE', 'UTF8'))

# get args from cgi
args = cgi.FieldStorage()
street = args['streetid'].value # 00732
nr = args['nr'].value # 10b

# Fetch dates from AHA
r=requests.post("http://www.aha-region.de/abfuhrkalender.html", data={'gemeinde': 'Hannover', 'strasse': street, 'hausnr': nr})

# Parse HTML page
result = {}
root = fromstring(r.text)
tbody=root.xpath('//div[@class="tx-vcdrabfallkalender-pi1"]/table/tbody')[0]
for tr in tbody.getchildren():
  if tr[0].tag == 'th':
    continue # skip table header
  if len(tr) < 5:
    continue
  what = tr[1].text
  for dat in tr[3].itertext():
    if '###' in dat:
      continue # skip comments
    dat = dat.strip().strip('*').strip()
    dat = datetime.strptime(dat,'%a, %d.%m.%Y')
    if not dat in result:
      result[dat] = []
    if not what in result[dat]:
      result[dat].append(what)

# Create Calendar
cal = vobject.iCalendar()
cal.add('method').value = 'PUBLISH' # IE/Outlook needs this
cal.add('X-WR-CALNAME').value = "AHA Muellabfuhrtermine"
cal.add('X-WR-TIMEZONE').value = time.tzname[time.localtime().tm_isdst]
for day, items in result.iteritems():
    vevent = cal.add('vevent')
    val = ", ".join(items)
    vevent.add('summary').value = val
    vevent.add('dtstart').value = day.date()
    vevent.add('dtend').value = (day+timedelta(days=1)).date()
    vevent.add('uid').value = val+str(day.date())
icalstream = cal.serialize()


print "Content-Type: text/calendar"
print ""
print icalstream

html_aha

#!/bin/bash 

. ../venv/bin/activate
python ../generate.py

requirements.txt

BeautifulSoup==3.1.0.1
lxml==2.2.8
requests==1.1.0
vobject==0.8.1c