def create_pt(delt_name=None, create_amount=100, pre_dict=None):
logger.info('delt_name:%s, create_amount:%d' % (delt_name, create_amount))
if create_amount <= 0:
return []
if delt_name:
key, country, roas_average_floor = mongo_prod.get_key_country(
delt_name)
logger.info('the platform: %s, the country: %s' % (key, country))
if key is None or country is None:
return []
else:
return []
global_vars = DeliveryVars()
global_vars.ads = pre_dict['ads']
if len(global_vars.ads) < 1:
return []
ads_count = dict()
ads_other = dict()
for ad_id in global_vars.ads:
ads_other[ad_id] = global_vars.ads[ad_id]['coef']
if global_vars.ads[ad_id]['platform'] == key:
ads_count[ad_id] = global_vars.ads[ad_id]['coef']
if len(ads_count) == 0:
return []
global_vars.ads_count = common_prod.weighted(ads_count)
global_vars.ads_other = ads_other
if len(ads_count) == 0:
return []
pt_pool = list()
while True:
''' 根据给定投放名称的平台随机获取一个父样本 '''
father_ad_id = choice(list(global_vars.ads_count.keys()),
1,
p=list(global_vars.ads_count.values()))[0]
copy_other = copy.deepcopy(global_vars.ads_other)
''' 在全部中删除获取的父样本，然后再剩下的样本中随机选取一个母样本'''
del copy_other[father_ad_id]
if len(copy_other) == 0:
break
copy_other = common_prod.weighted(copy_other)
mother_ad_id = choice(list(copy_other.keys()),
1,
p=list(copy_other.values()))[0]
pt_out = compose_baby(country, global_vars.ads[father_ad_id],
global_vars.ads[mother_ad_id], delt_name,
roas_average_floor, pre_dict)
pt_pool.extend(pt_out)
if len(pt_pool) >= create_amount:
break
mongo_prod.insert_baits(pt_pool)
mongo_prod.close()
logger.info('create amount is %d' % (len(pt_pool)))
return {'size': len(pt_pool)}
def choice_interest(father_pt, mother_pt, all_name, all_weight):
interests = list()
if len(all_name) == 0:
return interests
tmp_all_weight = copy.deepcopy(all_weight)
fid = dict()
mid = dict()
tmp_ids = []
if common_prod.node_exist(father_pt, 'interests'):
finterests = father_pt['adset_spec']['targeting']['interests']
if isinstance(finterests, list):
for fi in finterests:
fid[fi['id']] = 1
elif isinstance(finterests, dict):
finterests = finterests.values()
for fi in finterests.items():
fid[fi['id']] = 1
if common_prod.node_exist(mother_pt, 'interests'):
minterests = mother_pt['adset_spec']['targeting']['interests']
if isinstance(minterests, list):
for mi in minterests:
mid[mi['id']] = 1
elif isinstance(minterests, dict):
minterests = minterests.values()
for mi in minterests.items():
mid[mi['id']] = 1
mid_weight = all_weight[list(tmp_all_weight.keys())[int(len(all_weight)/2)]]
for fk in list(fid.keys()):
if all_weight.get(fk):
if mid.get(fk):
interests.append({'id': fk, 'name': all_name[fk]})
del mid[fk]
elif all_weight[fk] >= mid_weight:
interests.append({'id': fk, 'name': all_name[fk]})
tmp_ids.append(fk)
for mk in mid.keys():
if all_weight.get(mk):
if all_weight[mk] >= mid_weight and len(interests) < 400:
interests.append({'id': mk, 'name': all_name[mk]})
tmp_ids.append(mk)
rand_size = round((400 - len(interests))*random.uniform(0.1, 0.4))
''' 随机一个[0.1,0.5],然后乘以已存在interests的量和400的差值，再作新生成的随机产生的量 '''
if rand_size > 0:
for tpid in tmp_ids:
del tmp_all_weight[tpid]
prob = list(common_prod.weighted(tmp_all_weight).values())
rand_ids = choice(list(tmp_all_weight.keys()), rand_size, p=prob, replace=False)
if rand_size <= len(tmp_all_weight):
for rid in rand_ids:
interests.append({'id': round(float(rid), 1), 'name': all_name[rid]})
return interests
def choice_geolocation(geo_name=None, geo_weight=None):
geo_location = list()
if len(geo_name) > 0:
tmp_all_weight = common_prod.weighted(geo_weight)
rand_size = random.randint(10, 160)
rand_ids = choice(list(tmp_all_weight.keys()), rand_size, p=list(tmp_all_weight.values()), replace=False)
for rids in rand_ids:
rid = rids.split('_')
if len(rid) == 3:
geo_location.append({'latitude': rid[0], 'longitude': rid[1],
'radius': rid[2], 'distance_unit': geo_name[rids]})
return geo_location
def create_pt(delt_name=None, create_amount=10):
logger.info('delt_name:%s, create_amount:%d' % (delt_name, create_amount))
if create_amount <= 0:
return []
if delt_name:
key, country = mongo_prod.get_key_country(delt_name)
logger.info('the platform: %s, the country: %s' % (key, country))
if key is None or country is None:
return []
else:
return []
global_vars = DeliveryVars()
global_vars.ads = mongo_prod.select_ads(create_amount)
if len(global_vars.ads) < 1:
return []
ads_count = dict()
ads_other = dict()
for ad_id in global_vars.ads:
ads_other[ad_id] = global_vars.ads[ad_id]['coef']
if global_vars.ads[ad_id]['platform'] == key:
ads_count[ad_id] = global_vars.ads[ad_id]['coef']
if len(ads_count) == 0:
return []
global_vars.ads_count = common_prod.weighted(ads_count)
global_vars.ads_other = ads_other
if len(ads_count) == 0:
return []
''' 获取interests信息的属性值名称以及权重 '''
all_name, all_weight = mysql_prod.select_weigth()
''' 根据国家获取坐标信息 '''
geo_name, geo_weight = mysql_prod.select_geo_weigth(country)
creative_media = mysql_prod.select_url()
pts_size = 0
while True:
''' 根据给定投放名称的平台随机获取一个父样本 '''
father_ad_id = choice(list(global_vars.ads_count.keys()),
1,
p=list(global_vars.ads_count.values()))[0]
copy_other = copy.deepcopy(global_vars.ads_other)
''' 在全部中删除获取的父样本，然后再剩下的样本中随机选取一个母样本'''
del copy_other[father_ad_id]
if len(copy_other) == 0:
break
copy_other = common_prod.weighted(copy_other)
mother_ad_id = choice(list(copy_other.keys()),
1,
p=list(copy_other.values()))[0]
pt_out = compose_baby(country, global_vars.ads[father_ad_id],
global_vars.ads[mother_ad_id], all_name,
all_weight, geo_name, geo_weight, delt_name,
creative_media)
pts_size += mongo_prod.insert_baits(pt_out)
if pts_size >= create_amount:
break
mongo_prod.close()
logger.info('create amount is %d' % (pts_size))
return {'size': pts_size}
def create_pt(delt_name=None, create_amount=10):
# print(delt_name, create_amount)
logger.info('start create pt, two parameters are delt_name:%s, create_amount:%d' % (delt_name, create_amount))
if create_amount <= 0:
return []
if delt_name:
key, country = mongo_prod.get_key_country(delt_name)
logger.info('the platform: %s, the country: %s' % (key, country))
if key is None or country is None:
return []
else:
return []
global_vars = DeliveryVars()
cur_len = mongo_prod.get_income_logs_len()
if global_vars.cur_income_logs_len != cur_len:
global_vars.cur_income_logs_len = cur_len
global_vars.ads = mongo_prod.select_ads(create_amount)
logger.info('the size of global_vars.ads is %d' % (len(global_vars.ads)))
if len(global_vars.ads) == 0:
return []
ads_count = dict()
ads_other = dict()
for ad_id in global_vars.ads:
ads_other[ad_id] = global_vars.ads[ad_id]['coef']
if global_vars.ads[ad_id]['platform'] == key:
ads_count[ad_id] = global_vars.ads[ad_id]['coef']
if len(ads_count) == 0:
return []
global_vars.ads_count = common_prod.weighted(ads_count)
global_vars.ads_other = ads_other
logger.info('the size of same platform is %d, other platform is %d' % (len(ads_count), len(ads_other)))
if len(ads_count) == 0:
return []
pt_pool = list()
''' 获取interests信息的属性值名称以及权重 '''
all_name, all_weight = mysql_prod.select_weigth()
logger.info('the size of interests in dw_dim_interests is %d' % (len(all_name)))
''' 根据国家获取坐标信息 '''
geo_name, geo_weight = mysql_prod.select_geo_weigth(country)
logger.info('the size of coordinates in dw_dim_coordinate is %d' % (len(geo_name)))
creative_media = mysql_prod.select_url()
logger.info('the size of creative_media is %d' % len(creative_media))
while True:
''' 根据给定投放名称的平台随机获取一个父样本 '''
father_ad_id = choice(list(global_vars.ads_count.keys()), 1, p=list(global_vars.ads_count.values()))[0]
copy_other = copy.deepcopy(global_vars.ads_other)
''' 在全部中删除获取的父样本，然后再剩下的样本中随机选取一个母样本'''
del copy_other[father_ad_id]
if len(copy_other) == 0:
break
copy_other = common_prod.weighted(copy_other)
mother_ad_id = choice(list(copy_other.keys()), 1, p=list(copy_other.values()))[0]
pt_out = compose_baby(country, global_vars.ads[father_ad_id], global_vars.ads[mother_ad_id], all_name,
all_weight, geo_name, geo_weight, delt_name, creative_media)
for pp in pt_out:
pt_pool.append(pp)
if len(pt_pool) >= create_amount:
break
mongo_prod.close()
logger.info('create amount pt is %d' % (len(pt_pool)))
return pt_pool
