Browse Source

Add Medway scraper.

master
duncan.parkes 16 years ago
parent
commit
442f48ab30
4 changed files with 93 additions and 0 deletions
  1. +1
    -0
      python_scrapers/Maldon.py
  2. +90
    -0
      python_scrapers/Medway.py
  3. +1
    -0
      python_scrapers/OtherFilesToCopy.csv
  4. +1
    -0
      python_scrapers/SitesToGenerate.csv

+ 1
- 0
python_scrapers/Maldon.py View File

@@ -97,3 +97,4 @@ if __name__ == '__main__':

# 1) Check that it works ok on a no results page.
# 2) Email the council about non-linkable info page.
# 3) Email the council about missing descriptions?

+ 90
- 0
python_scrapers/Medway.py View File

@@ -0,0 +1,90 @@
import urllib2
import urllib
import urlparse

import datetime, time

from BeautifulSoup import BeautifulSoup

from PlanningUtils import PlanningApplication, \
PlanningAuthorityResults, \
getPostcodeFromText

date_format = "%d/%m/%Y"

class MedwayParser:
comment_email_address = "planning.representations@medway.gov.uk"

def __init__(self, *args):
self.authority_name = "Medway Council"
self.authority_short_name = "Medway"

self.base_url = "http://www.medway.gov.uk/index/environment/planning/planapp/planonline.htm"
self._split_base_url = urlparse.urlsplit(self.base_url)

self._results = PlanningAuthorityResults(self.authority_name, self.authority_short_name)


def getResultsByDayMonthYear(self, day, month, year):
search_date = datetime.date(year, month, day)
search_date_string = search_date.strftime(date_format)

"appstat=&decision=&appdec=&ward=&parish=&dadfrom=&dadto=&davfrom=01%2F06%2F2008&davto=02%2F06%2F2008&searchbut=Search"
search_data = urllib.urlencode(
[("searchtype", "1"),
("appstat", ""),
("decision", ""),
("appdec", ""),
("ward", ""),
("parish", ""),
("dadfrom", ""),
("dadto", ""),
("davfrom", search_date_string),
("davto", search_date_string),
("searchbut", "Search"),
]
)

split_search_url = self._split_base_url[:3] + (search_data, '')
search_url = urlparse.urlunsplit(split_search_url)

response = urllib2.urlopen(search_url)
soup = BeautifulSoup(response.read())

results_table = soup.find(text="Application No").parent.parent.parent
trs = results_table.findAll("tr")[1:]

tr_counter = 0
while tr_counter < len(trs):
tr = trs[tr_counter]

if tr_counter % 2 == 0:
application = PlanningApplication()
application.date_received = search_date
application.comment_url = self.comment_email_address

tds = tr.findAll("td")

application.info_url = urlparse.urljoin(self.base_url, tr.a['href'])
application.council_reference = tr.a.string.strip()

application.address = tds[1].string.strip()
application.postcode = getPostcodeFromText(application.address)

application.description = tds[2].string.strip()

self._results.addApplication(application)

tr_counter += 1

return self._results


def getResults(self, day, month, year):
return self.getResultsByDayMonthYear(int(day), int(month), int(year)).displayXML()

if __name__ == '__main__':
parser = MedwayParser()
print parser.getResults(02,6,2008)


+ 1
- 0
python_scrapers/OtherFilesToCopy.csv View File

@@ -29,3 +29,4 @@
"ForestOfDean.py", "420"
"Flintshire.py", "420"
"Maldon.py", "420"
"Medway.py", "420"

+ 1
- 0
python_scrapers/SitesToGenerate.csv View File

@@ -229,3 +229,4 @@
"Forest of Dean District Council", "Forest of Dean", "", "ForestOfDean", "ForestOfDeanParser"
"Flintshire County Council", "Flintshire", "", "Flintshire", "FlintshireParser"
"Maldon District Council", "Maldon", "", "Maldon", "MaldonParser"
"Medway Council", "Medway", "", "Medway", "MedwayParser"

Loading…
Cancel
Save