Geo-matching for Prediction CSVs

geomatch.py

# Go through a CSV file of addresses
# Find each address in a KML file
# Add the coordinates to the CSV file if a match is found

# Add Geostuff to addresses

cases = open('PROP1Cleanest.csv', 'r')
geocases = open('allcasesplusgeo.csv', 'w')
lastgeo = [ ]
geobyaddress = { }

for case in cases:
  myinfo = case.split('"F"')[1].split(",")
  myaddress = case.split('"')[1]
  if(geobyaddress.has_key(myaddress)):
    lastgeo = geobyaddress[ myaddress ]
    case = case.replace('"F","",""','"F",' + lastgeo[0] + "," + lastgeo[1])
    print lastgeo

  else:
    # new address, look up location
    lastgeo = [ ]
    coordinatesOn = 0
    caseid = myinfo[4]
    try:
      if(len(caseid) < 4 or caseid.find('"') > -1):
        # has an invalid caseid
        print "error: " + caseid
      geosearch = open('../Downloads/cases-' + caseid[0:4] + 'geo.kml', 'r')
      for geoline in geosearch:
        if(geoline.find('<b>CaseID:</b> ' + caseid + '<br>') > -1):
          # found this address
          coordinatesOn = 1
        
        if(geoline.find('</Placemark>') > -1 and coordinatesOn == 1):
          # no geocode exists for this address
          break

        if(geoline.find('<coordinates>') > -1 and coordinatesOn == 1):
          # found the coordinates
          # set case[5] to latitude, case[6] to longitude
          lat = geoline.split(',')[0]
          lat = lat[lat.find('coordinates') + 12: len(lat)]
          lng = geoline.split(',')[1]
          lastgeo = [ lat, lng ]
          geobyaddress[ myaddress ] = lastgeo
          case = case.replace('"F","",""','"F",' + lat + "," + lng)
          break

      # didn't find caseid in expected file
      # check next year as a back-up
      geosearch = open('../Downloads/cases-' + str(int(caseid[0:4]) + 1) + 'geo.kml', 'r')
      for geoline in geosearch:
        if(geoline.find('<b>CaseID:</b> ' + caseid + '<br>') > -1):
          # found this address
          coordinatesOn = 1
        
        if(geoline.find('</Placemark>') > -1 and coordinatesOn == 1):
          # no geocode exists for this address
          break

        if(geoline.find('<coordinates>') > -1 and coordinatesOn == 1):
          # found the coordinates
          # set case[5] to latitude, case[6] to longitude
          lat = geoline.split(',')[0]
          lat = lat[lat.find('coordinates') + 12: len(lat)]
          lng = geoline.split(',')[1]
          lastgeo = [ lat, lng ]
          geobyaddress[ myaddress ] = lastgeo
          case = case.replace('"F","",""','"F",' + lat + "," + lng)
          break
      
    except:
      # caseid was not valid year
      continue

  if(lastgeo != [ ]):
    # print the geocoded case
    geocases.write(case)