def group_ungrouped_rsts(city_id):
""" Group all ungrouped restaurants into their nearly regions
:param city_id:
:return:
"""
ungrouped_rsts = get_ungrouped_rst(city_id)
regions = region_base.get_regions_by_city_id(city_id)
region_infos = [[r['id'], region_area_to_polygons(r['area'])]
for r in regions]
for region_info in region_infos:
region_info.append(get_centerpoint_of_region(region_info[1]))
if not ungrouped_rsts:
return 0
for rst in ungrouped_rsts:
point = (float(rst.latitude), float(rst.longitude))
region_id = None
for region in regions:
if not region['area']:
continue
polygons = region_area_to_polygons(region['area'])
for polygon in polygons:
if geo.is_in_region(point, polygon, xy_reverse=True):
region_id = region['id']
break
if region_id:
break
if not region_id:
region_id = region_infos[0]
min_distance = sys.maxint
for info in region_infos:
center_points = info[2]
cur_distance = min([get_dist(p, point) for p in center_points])
if cur_distance < min_distance:
min_distance = cur_distance
region_id = info[0]
log.info('Restaurant {} does not have matched regions. Find closest'
' one with region_id {}, distance {}'.format(
rst.id, region_id, min_distance))
log.info('update rst {} with region {}'.format(rst.id, region_id))
inner.update_restaurant_region(rst.id, region_id)
return len(ungrouped_rsts)
def group_all_restaurants(city_id, center_point, max_distance_to_center,
max_region_rsts_num):
# for logging
rst_in_region_num = 0
rst_near_region_num = 0
rst_far_from_region_num = 0
exc_num = 0
# 1. get all valid and not BOD restaurants
rsts = inner.query_all_rsts(city_ids=[city_id], is_valid=1, is_premium=0,
region_type=region_base.REGION_TYPE_CBD)
rsts_info = [{'id': r['id'],
'lng': r['longitude'],
'lat': r['latitude']} for r in rsts]
rst_len = len(rsts)
log.info('Step 1: query. all rsts to group count is {}'.format(rst_len))
# 2. sort restaurants by their distance to center of the city
for rst in rsts_info:
rst['center_dist'] = geo.distance(center_point,
(rst['lng'], rst['lat']))
rst['min_region_dist'] = rst['region_id'] = None
sorted_rsts = sorted(rsts_info, key=lambda rest: rest['center_dist'])
log.info('Step 2: sorted complete, farest dist is {}'.
format(sorted_rsts[-1]['center_dist']))
# 3. calculate each region's center point and radius
regions = region_inner.query_regions(city_ids=[city_id],
type_codes=[REGION_TYPE_CBD])
regions_info = {r['id']:
{'area': region_area_to_polygons(r['area'], reverse=True)}
for r in regions}
region_id_to_del = []
for region_id, region in regions_info.iteritems():
region['center'] = get_centerpoint_of_region(region['area'], only_one=True)
region['radius'] = get_radius_of_region(region['center'], region['area'])
# info['rst_ids'] = []
region['rst_num'] = 0
if not (region['center'] and region['radius']):
region_id_to_del.append(region_id)
for r_id in region_id_to_del:
regions_info.__delitem__(r_id)
log.info('Step 3: calculate center and radius')
# 4. main loop for restaurants, there are three kinds of results
# for each restaurant:
# a. in certain region
# b. near certain region (no more than `max_distance_to_center`)
# c. others (drop into ungrouped region)
for index, rst in enumerate(sorted_rsts):
try:
if index % 100 == 0:
log.info('grouping rate {}%'.format(float(index) / rst_len * 100))
rst_point = (rst['lng'], rst['lat'])
final_region_id = None
# a. if in certain region
in_region_flag = False
for region_id, region in regions_info.iteritems():
if not region['area']:
continue
polygons = region['area']
for polygon in polygons:
if geo.is_in_region(rst_point, polygon, xy_reverse=True):
in_region_flag = True
if region['rst_num'] < max_region_rsts_num:
rst['region_id'] = region_id
region['rst_num'] += 1
rst_in_region_num += 1 # for logging
break
if in_region_flag:
break
if rst.get('region_id'):
continue
# b. if near some region
min_dist = sys.maxint
for region_id, region in regions_info.iteritems():
current_dist = geo.distance(rst_point, region['center'])
if current_dist < min_dist and \
region['rst_num'] < max_region_rsts_num:
min_dist = current_dist
if min_dist < max_distance_to_center + region['radius']:
rst['min_region_dist'] = min_dist
region['rst_num'] += 1
final_region_id = region_id
rst_near_region_num += 1 # for logging
break
if not final_region_id:
rst['region_id'] = None
rst_far_from_region_num += 1 # for logging
except Exception as e:
exc_num += 1
print(e)
rst_total_num = len(sorted_rsts)
average_rst_region_dist = 0
rst_count = 0
for rst in sorted_rsts:
if rst['min_region_dist'] is not None:
rst_count += 1
average_rst_region_dist += rst['min_region_dist']
average_rst_region_dist /= rst_count
region_average_rst_num = 0
region_rst_num_count = 0
region_average_radius = 0
region_radius_count = 0
for region_id, region in regions_info.iteritems():
if region['rst_num'] is not None:
region_rst_num_count += 1
region_average_rst_num += region['rst_num']
if region['radius'] is not None:
region_radius_count += 1
region_average_radius += region['radius']
region_average_rst_num /= region_rst_num_count
region_average_radius /= region_radius_count
log.info('Step 4: Main grouping. total_rst_num: {}, total_region_num: {}, '
'rst_in_region: {}, rst_near_region: {}, other_rsts: {};'
'ave rst region dist: {}, ave region rst num: {},'
' ave region radius {}. total exc {}.'.
format(rst_total_num, len(regions_info), rst_in_region_num,
rst_near_region_num, rst_far_from_region_num,
average_rst_region_dist, region_average_rst_num,
region_average_radius, exc_num))
# 5. update restaurant region
for index, rst in enumerate(sorted_rsts):
try:
if index % 200 == 0:
log.info('updating rate {}%'.format(float(index) / rst_len * 100))
# log.info('update rst {} with region {}'.format(rst.id, region_id))
# inner.update_restaurant_region(rst['id'], rst['region_id'])
log.info('rst_id: {}, region_id: {}'.format(rst['id'], rst['region_id']))
except Exception as e:
exc_num += 1
print(e)
for region_id, region in regions_info.iteritems():
log.info('{}: {}'.format(region_id, region['rst_num']))
log.info('done.')
return True
