Automatically exported from code.google.com/p/planningalerts
您最多选择25个主题 主题必须以字母或数字开头,可以包含连字符 (-),并且长度不得超过35个字符
 
 
 
 
 
 

101 行
3.4 KiB

  1. import urllib2
  2. import urllib
  3. import urlparse
  4. import datetime, time
  5. import cgi
  6. import re
  7. import cookielib
  8. cookie_jar = cookielib.CookieJar()
  9. from BeautifulSoup import BeautifulSoup
  10. from PlanningUtils import PlanningApplication, \
  11. PlanningAuthorityResults, \
  12. getPostcodeFromText
  13. from HTTPHandlers import CookieAddingHTTPRedirectHandler
  14. cookie_handling_opener = urllib2.build_opener(CookieAddingHTTPRedirectHandler(cookie_jar))
  15. search_date_format = "%m/%d/%Y" #That's right, the search date is US style.
  16. info_page_date_format = "%d/%m/%Y" # and the info page is UK style
  17. class GosportParser:
  18. def __init__(self, *args):
  19. self.authority_name = "Gosport Borough Council"
  20. self.authority_short_name = "Gosport"
  21. self.base_url = "http://www.gosport.gov.uk/gbcplanning/ApplicationSearch2.aspx"
  22. self.info_url = "http://www.gosport.gov.uk/gbcplanning/ApplicationDetails.aspx?ID=%s"
  23. self._results = PlanningAuthorityResults(self.authority_name, self.authority_short_name)
  24. def getResultsByDayMonthYear(self, day, month, year):
  25. search_date = datetime.date(year, month, day)
  26. get_request = urllib2.Request(self.base_url)
  27. get_response = urllib2.urlopen(get_request)
  28. cookie_jar.extract_cookies(get_response, get_request)
  29. get_soup = BeautifulSoup(get_response.read())
  30. post_data = (
  31. ("__VIEWSTATE", get_soup.find("input", {"name": "__VIEWSTATE"})["value"]),
  32. ("pgid", get_soup.find("input", {"name": "pgid"})["value"]),
  33. ("action", "Search"),
  34. # ("ApplicationSearch21%3AtbDevAddress", ""),
  35. # ("ApplicationSearch21%3AtbApplicantName", ""),
  36. # ("ApplicationSearch21%3AtbAgentName", ""),
  37. ("ApplicationSearch21:tbDateSubmitted", search_date.strftime(search_date_format)),
  38. ("ApplicationSearch21:btnDateSubmitted", "Search"),
  39. # ("ApplicationSearch21%3AtbDateDetermined", ""),
  40. )
  41. post_request = urllib2.Request(self.base_url, urllib.urlencode(post_data))
  42. cookie_jar.add_cookie_header(post_request)
  43. post_response = cookie_handling_opener.open(post_request)
  44. post_soup = BeautifulSoup(post_response.read())
  45. # Discard the first <tr>, which contains headers
  46. trs = post_soup.find("table", id="SearchResults1_dgSearchResults").findAll("tr")[1:]
  47. for tr in trs:
  48. application = PlanningApplication()
  49. tds = tr.findAll("td")
  50. application.council_reference = tds[0].string.strip()
  51. application.address = tds[1].string.strip()
  52. application.postcode = getPostcodeFromText(application.address)
  53. application.description = tds[2].string.strip()
  54. application.date_received = datetime.datetime(*(time.strptime(tds[3].string.strip(), info_page_date_format)[0:6]))
  55. application.info_url = self.info_url %(application.council_reference)
  56. # The comment url must be accessed by a POST, so we'll just use the info url for that as well
  57. application.comment_url = application.info_url
  58. self._results.addApplication(application)
  59. return self._results
  60. def getResults(self, day, month, year):
  61. return self.getResultsByDayMonthYear(int(day), int(month), int(year)).displayXML()
  62. if __name__ == '__main__':
  63. parser = GosportParser()
  64. print parser.getResults(12,6,2009)