Scrape KML name ans boundaries out of mapitmysociety suitable for import to fusion tables. Currently set to scrape wards from the LB of Tower Hamlets

boundary_scraper.py

import scraperwiki
import simplejson
import urllib2
import csv
from lxml import etree

# the borough key from mapitmysociety URL for the borough page
TYP = 'LBW'
KEY = '2506'
# typs are:
'''CTY (county council), CED (county ward), COI (Isles of Scilly),
COP (Isles of Scilly parish), CPC (civil parish/community),
CPW (civil parish/community ward), DIS (district council),
DIW (district ward), EUR (Euro region), GLA (London Assembly),
LAC (London Assembly constituency), LBO (London borough),
LBW (London ward), LGD (NI council), LGE (NI electoral area),
LGW (NI ward), MTD (Metropolitan district), MTW (Metropolitan ward),
NIE (NI Assembly constituency), OLF (Lower Layer Super Output Area, Full),
OLG (Lower Layer Super Output Area, Generalised),
OMF (Middle Layer Super Output Area, Full),
OMG (Middle Layer Super Output Area, Generalised),
SPC (Scottish Parliament constituency),
SPE (Scottish Parliament region), UTA (Unitary authority),
UTE (Unitary authority electoral division), UTW (Unitary authority ward),
WAC (Welsh Assembly constituency), WAE (Welsh Assembly region),
WMC (UK Parliamentary constituency)'''


csvfile = open(str(KEY) + 'ward_data.csv', 'w')
csvwriter = csv.writer(csvfile)
csvwriter.writerow(['name','geomtery']) 
#Get a stub KML file for the local council level
url='http://mapit.mysociety.org/area/'+str(KEY)+'.kml'
xmlraw = urllib2.urlopen(url).read()
xml=etree.fromstring(xmlraw)
 
#Get the list of electoral wards covered by that council area
wards=simplejson.load(urllib2.urlopen('http://mapit.mysociety.org/area/'+str(KEY)+'/covers?type='+TYP))
 
#Get the name and Polgon for each ward, stick them in a CSV
for ward in wards:
    url='http://mapit.mysociety.org/area/'+ward+'.kml'
    xmlraw = scraperwiki.scrape(url)
    xml2=etree.fromstring(xmlraw)
    # print etree.tostring(xml2)
    names = xml2.xpath('//kml:name', namespaces={'kml': "http://www.opengis.net/kml/2.2"})
    poly = xml2.xpath('//kml:Polygon', namespaces={'kml': "http://www.opengis.net/kml/2.2"})
    print names[0].text
    csvwriter.writerow([names[0].text,etree.tostring(poly[0])])