def load_licenses(filename, property):
"""Load in business license data."""
businesses = {}
with open(filename) as stream:
reader = data.RawReader(stream)
for row in reader:
neighborhood = row.neighborhood
if property == 'start_date':
year = row.start_date.year
elif property == 'end_date':
if row.end_date:
year = row.end_date.year
else:
year = this_year
account_number = row.account_number
site_number = row.site_number
businesses[(account_number, site_number)] = {
'neighborhood': neighborhood,
'year': year
}
return businesses
import data
# read shapefile data using fiona
boundaries = {}
with fiona.drivers():
with fiona.open(sys.argv[1]) as source:
neighborhood_projection = pyproj.Proj(preserve_units=True, **source.crs)
for thing in source:
neighborhood = thing['properties']['PRI_NEIGH']
boundaries[neighborhood] = shape(thing['geometry'])
# add the neighborhood to the row using a brute force search, which could be
# sped up computationally if necessary
reader = data.RawReader(sys.stdin)
rows = []
for row in reader:
rows.append(row)
row['NEIGHBORHOOD'] = ''
if row['LONGITUDE'] and row['LATITUDE']:
x, y = map(float, [row[k] for k in ['LONGITUDE', 'LATITUDE']])
point = Point(neighborhood_projection(x, y))
neighborhoods = set()
for neighborhood, geometry in boundaries.iteritems():
if geometry.contains(point):
neighborhoods.add(neighborhood)
assert len(neighborhoods) <= 1, "%s\n\n%s" % (row, neighborhoods)
if neighborhoods:
row['NEIGHBORHOOD'] = neighborhoods.pop()
class Counter(collections.Counter):
year_range = range(2004, 2015)
def restrict_to_year_range(self, multiplier=1):
output = []
for year in self.year_range:
output.append(multiplier * self[year])
return output
out_dir = sys.argv[-1]
with open(sys.argv[1]) as stream:
reader = data.RawReader(stream)
neighborhood = None
new_counts, old_counts = Counter(), Counter()
for row in reader:
year = int(row['year'])
if neighborhood is None:
neighborhood = row['neighborhood']
if neighborhood != row['neighborhood']:
if not neighborhood:
neighborhood = "unknown"
fig = figs.FlowOverTime(
Counter.year_range,
new_counts.restrict_to_year_range(),
old_counts.restrict_to_year_range(multiplier=-1),
)
filename = os.path.join(
