Automatically exported from code.google.com/p/planningalerts
Você não pode selecionar mais de 25 tópicos Os tópicos devem começar com uma letra ou um número, podem incluir traços ('-') e podem ter até 35 caracteres.
 
 
 
 
 
 

79 linhas
2.9 KiB

  1. import urllib2
  2. import urllib
  3. import urlparse
  4. import datetime
  5. import cgi
  6. from BeautifulSoup import BeautifulSoup
  7. from PlanningUtils import PlanningApplication, \
  8. PlanningAuthorityResults, \
  9. getPostcodeFromText
  10. class CarmarthenshireParser:
  11. def __init__(self, *args):
  12. self.comments_email_address = "planning@carmarthenshire.gov.uk"
  13. self.authority_name = "Carmarthenshire County Council"
  14. self.authority_short_name = "Carmarthenshire"
  15. self.base_url = "http://www.carmarthenshire.gov.uk/CCC_APPS/eng/plannaps/CCC_PlanningApplicationsResults.asp?datemode=range&in_lo_date=%(day)s%%2F%(month)s%%2F%(year)s&in_hi_date=%(day)s%%2F%(month)s%%2F%(year)s&SUBMIT=Search"
  16. self._results = PlanningAuthorityResults(self.authority_name, self.authority_short_name)
  17. def getResultsByDayMonthYear(self, day, month, year):
  18. search_day = datetime.date(year, month, day)
  19. # Now get the search page
  20. response = urllib2.urlopen(self.base_url %{"day": day,
  21. "month": month,
  22. "year": year,
  23. })
  24. soup = BeautifulSoup(response.read())
  25. trs = soup.findAll("tr", valign="middle")
  26. count = 0
  27. for tr in trs:
  28. # The odd trs are just spacers
  29. if count % 2 == 0:
  30. application = PlanningApplication()
  31. tds = tr.findAll("td")
  32. application.date_received = search_day
  33. application.council_reference = tds[1].a.string
  34. application.address = tds[3].a.string
  35. application.postcode = getPostcodeFromText(application.address)
  36. # All the links in this <tr> go to the same place...
  37. application.info_url = urlparse.urljoin(self.base_url, tr.a['href'])
  38. # Still looking for description and comment url
  39. # For the description, we'll need the info page
  40. info_soup = BeautifulSoup(urllib2.urlopen(application.info_url).read())
  41. application.description = info_soup.find(text="Description").findNext("td").findNext("td").font.string
  42. # While we're here, lets get the OSGB grid ref
  43. application.osgb_x, application.osgb_y = info_soup.find(text="Grid Reference").findNext("td").font.string.split("-")
  44. # We'll have to use an email address for comments
  45. application.comment_url = self.comments_email_address
  46. self._results.addApplication(application)
  47. count += 1
  48. return self._results
  49. def getResults(self, day, month, year):
  50. return self.getResultsByDayMonthYear(int(day), int(month), int(year)).displayXML()
  51. if __name__ == '__main__':
  52. parser = CarmarthenshireParser()
  53. print parser.getResults(8,8,2008)