def geocode_address(self, origin_address):
"""Geocodes an address string
Filters results by settings.STATE
Sets self.lat/lng to result
Throws exceptions if the number of results isn't exactly 1
"""
self.places = []
b = geocoders.Bing(settings.BING_KEY)
try:
place, (lat, lng) = b.geocode(origin_address)
self.places.append((place, lat, lng))
except ValueError:
for result in b.geocode(origin_address, exactly_one=False):
place, (lat, lng) = result
self.places.append((place, lat, lng))
except:
self.places.append(None)
self.places = _filter_addresses(self.places)
if len(self.places) > 1:
raise ValueError("More than one location found")
if len(self.places) == 0:
raise LookupError("No Locations in PA found")
else:
self.lat = self.places[0][1]
self.lng = self.places[0][2]
def read_scrub_list_and_update(args):
try:
filename = args[0]
except:
pass
# Thomas' Bing Geocoder api key (free basic access)
# dotus_geocoder = geocoders.GeocoderDotUS() for consideration as a fallback
bing_geocoder = geocoders.Bing(
'AvxLEwiPhVJzf0S3Pozgg01NnUQQX0RR6g9K46VPLlZ8OfZkKS-76gaPyzoV6IHI')
path = os.path.join(settings.BASE_DIR, 'readonly', filename)
try:
f = open(path)
except IOError:
print_stack_trace()
rows = []
for row in f:
rows.append(row.replace("\r\n", "").split("\t"))
for row in rows[1:]: # Skip the header
try:
coupon_pk = int(row[1])
is_duplicate = True if row[3] == '1' else False
is_inactive = True if row[4] == '1' else False
is_category_wrong = True if row[5] == '1' else False
is_location_wrong = True if row[8] == '1' else False
correction_needed = is_duplicate or is_inactive or is_category_wrong or is_location_wrong
if correction_needed:
coupon_obj = Coupon.all_objects.get(pk=coupon_pk)
if is_duplicate:
coupon_obj.is_duplicate = True
# print "Correction: ", coupon_pk, " is_duplicate=True" #DEBUG
if is_inactive:
coupon_obj.status = 'confirmed-inactive'
# print "Correction: ", coupon_pk, " status=confirmed-inactive" #DEBUG
if is_category_wrong:
coupon_obj.categories.clear()
try:
parent_category = Category.objects.get(
ref_id_source='sqoot', name=row[6])
coupon_obj.categories.add(parent_category)
# print "Correction: ", coupon_pk, " Parent category -> ", parent_category.name #DEBUG
except:
pass
try:
child_category = Category.objects.get(
ref_id_source='sqoot', name=row[7])
coupon_obj.categories.add(child_category)
# print "Correction: ", coupon_pk, " Child category -> ", child_category.name #DEBUG
except:
pass
if is_location_wrong:
location_obj = coupon_obj.merchant_location
address = row[9] if row[9] != '' else ''
locality = row[10] if row[10] != '' else ''
region = row[11] if row[11] != '' else ''
postal_code = row[12] if row[12] != '' else ''
spacer1 = ', ' if address != '' else ''
spacer2 = ' ' if locality != '' else ''
lookup_text = address + spacer1 + locality + spacer2 + region
try:
place, (lat, lng) = bing_geocoder.geocode(lookup_text)
pnt = 'POINT({} {})'.format(lng, lat)
location_obj.geometry = pnt
except:
pass
location_obj.address = address if address != '' else location_obj.address
location_obj.locality = locality if locality != '' else location_obj.locality
location_obj.region = region if region != '' else location_obj.region
location_obj.postal_code = postal_code if postal_code != '' else location_obj.postal_code
location_obj.save()
# print "Correction: ", coupon_pk, " Location fixed" #DEBUG
coupon_obj.save()
except:
print_stack_trace()
scrub_list_retrieved = [
row[1] for row in rows[1:]
] # list of original coupon pks imported from 'scrub_list.py'
deals_to_scrub = Coupon.all_objects.filter(pk__in=scrub_list_retrieved)\
.exclude(Q(status='confirmed-inactive') | Q(status='implied-inactive') | Q(is_duplicate=True))\
.order_by('merchant__name')
probably_dup_deals_list = [
] # List of coupon pks that look like a duplicate.
probably_dup_deals_list = crosscheck_by_field(deals_to_scrub,
probably_dup_deals_list,
'coupon_directlink')
probably_dup_deals_list = crosscheck_by_field(deals_to_scrub,
probably_dup_deals_list,
'merchant_name')
probably_dup_deals_list = list(set(probably_dup_deals_list))
for pk in probably_dup_deals_list:
try:
coupon = Coupon.all_objects.get(pk=pk)
coupon.is_duplicate = True
coupon.save()
# print "Correction: ", coupon_pk, " is_duplicate=True" #DEBUG
except:
print_stack_trace()
