def test_basic(self):
from pprint import pprint
from databundles.geo.geocoder import Geocoder
g = Geocoder(self.bundle.library)
filename = "good_segments"
f_input = os.path.join(os.path.dirname(__file__),'support',filename + '.txt')
f_output = os.path.join(os.path.dirname(__file__),'support',filename + '.out.csv')
with open(f_input) as f:
for line in f:
addr = line.strip()
r = g.geocode_address(addr)
print "==", addr
print "->",r
if r:
print " ", r['coded_address']
def x_test_crime(self):
from databundles.geo.address import Parser
from databundles.geo.geocoder import Geocoder
import csv
g = Geocoder(self.bundle.library, addresses_ds='geoaddresses')
_,incidents = self.bundle.library.dep('crime')
log_rate = self.bundle.init_log_rate(1000)
p = Parser()
with open(self.bundle.filesystem.path('errors.csv'), 'wb') as f:
writer = csv.writer(f)
writer.writerow(['code','arg','block_address','city','number','dir','street','type'])
multi_cities = 0.0
multi_addr = 0.0
no_response = 0.0
for i, inct in enumerate(incidents.query("SELECT * FROM incidents limit 100000")):
row = dict(inct)
candidates = g.geocode_semiblock(row['blockaddress'], row['city'], 'CA')
if len(candidates) == 0:
no_response += 1
self.write_error_row('norsp',0, p,writer,row['blockaddress'], row['city'])
continue
elif len(candidates) != 1:
multi_cities += 1
self.write_error_row('mcities',len(candidates), p,writer,row['blockaddress'], row['city'])
continue
s = candidates.popitem()[1]
if len(s) > 3:
self.write_error_row('maddr',len(s), p,writer,row['blockaddress'], row['city'])
multi_addr +=1
if i > 0:
log_rate("{} cities={}, {}% addr={}, {}% nrp={}, {}%".format(i,
multi_cities, int(multi_cities/i * 100),
multi_addr, int(multi_addr/i * 100),
no_response, int(no_response/i * 100) ))
def test_geo(self):
from databundles.geo.geocoder import Geocoder
g = Geocoder(self.library, addresses_ds='geoaddresses')
p = self.partitions.find(table='businesses')
errorp = self.partitions.find_or_new(table='businesses', grain='errors')
ok = 0
errors = 0
with errorp.database.inserter() as ins:
for row in p.query('SELECT * FROM businesses'):
candidates = g.geocode_address(row['address'], row['city'], 'CA')
if len(candidates) != 1 :
#print "('{0}', (None, '{0}','gln')),".format(row['address'])
errors += 1
else:
ok += 1
print len(candidates), ok, errors, int(float(ok)/(ok+errors) * 100)
ins.insert(row)
