def special_brand_loading(self):
ex_brand_dict = tool.get_exchange_brand_pair()
special_brand_list = []
special_brand_dict = {}
b_id_dict = {}
with open(self.special_brand_file, "r", encoding="utf-8") as f1:
for line in f1:
line = line.strip()
if line == "": continue
# brand_id, brand_name, cat1_id, cat1, gmv
lst1 = line.split("\t")
if len(lst1) != 2:
continue
lst1 = [tmp.strip() for tmp in lst1]
b_id, b_name = lst1
if b_id in b_id_dict:
continue
b_id_dict[b_id] = ''
if b_name in ex_brand_dict:
b_name = ex_brand_dict[b_name]
b_name = tool.brand_clean(b_name.lower())
ext_band_name = self.english_brand_extension(b_name)
special_brand_list.append(ext_band_name)
special_brand_dict[b_id] = b_name
return special_brand_dict, special_brand_list
def brand_recall_local_file(self):
try:
ex_brand_dict = self.brand_exchange_dict
r_lst = []
with open(self.in_file) as f1:
for line in f1:
line = line.strip()
if line == "": continue
lst1 = line.split("\t")
if len(lst1) != 5:
continue
lst1 = [tmp.strip() for tmp in lst1]
b_id, ori_b_name, cat1_name, cat2_name, cat3_name = lst1
if ori_b_name in self.brand_del_dict: continue
if b_id in self.del_brandID_dict: continue
if ori_b_name in ex_brand_dict:
b_name = ex_brand_dict[ori_b_name]
else:
b_name = ori_b_name
b_name = tool.brand_clean(b_name)
ext_band_name = self.english_brand_extension(b_name)
r_lst.append("\t".join([
b_id, ext_band_name, ori_b_name, cat1_name, cat2_name,
cat3_name
]))
with open(self.out_file, "w") as f1:
f1.write("\n".join(r_lst))
f1.flush()
except Exception as e:
raise e
def brand_recall(self, output_file="pdd_xjd_brand_recall_info.txt"):
ex_brand_dict = tool.get_exchange_brand_pair()
special_brand_list = []
mijia_lst = []
with open(self.standard_brand_file, "r", encoding="utf-8") as f1:
for line in f1:
line = line.strip()
if line == "": continue
# brand_id, brand_name, cat2_id, cat2, gmv
lst1 = line.split("\t")
if len(lst1) != 5:
continue
lst1 = [tmp.strip() for tmp in lst1]
b_id, b_name, cat2_id, cat2, gmv = lst1
if len(b_name) == 1: continue
if self.del_brand_id(b_id): continue
ori_b_name = b_name
if b_name in ex_brand_dict:
b_name = ex_brand_dict[b_name]
b_name = tool.brand_clean(b_name.lower())
ext_band_name = self.english_brand_extension(b_name)
special_brand_list.append("\t".join(
[b_id, ori_b_name, ext_band_name, cat2_id, cat2, gmv]))
s9 = self.mijia_special_brand_recall(b_id, cat2_id, cat2)
if s9 != "": mijia_lst.append(s9)
r_lst = special_brand_list + mijia_lst
with open(output_file, "w", encoding="utf-8") as f1:
f1.write("\n".join(r_lst))
f1.flush()
def special_brand_dealing(self, lst1, ex_brand_dict):
b_id, b_name = lst1
if b_name in ex_brand_dict:
b_name = ex_brand_dict[b_name]
b_name = tool.brand_clean(b_name.lower())
ext_band_name = self.english_brand_extension(b_name)
return ext_band_name
def _brand_ext(self, ori_b_name):
if ori_b_name in self.brand_exchange_dict:
b_name = self.brand_exchange_dict[ori_b_name]
else:
b_name = ori_b_name
b_name = b_name.strip()
if b_name == "": return ""
b_name = tool.brand_clean(b_name)
ext_band_name = self.english_brand_extension(b_name)
return ext_band_name
def brand_recall(self, output_file="brand_recall_info.txt"):
try:
ex_brand_dict = tool.get_exchange_brand_pair()
special_brand_list = self.special_brand_loading()
recall_brand_dict = {}
mijia_lst = []
idx = 0
with open(self.standard_brand_file, "r", encoding="utf-8") as f1:
for line in f1:
line = line.strip()
if line == "": continue
lst1 = line.split("\t")
if len(lst1) != 5:
continue
lst1 = [tmp.strip() for tmp in lst1]
b_id, ori_b_name, cat1_id, cat1, gmv = lst1
mijia_str = self.mijia_special_brand_recall(
b_id, cat1_id, cat1)
if mijia_str != "": mijia_lst.append(mijia_str)
if ori_b_name in ex_brand_dict:
b_name = ex_brand_dict[ori_b_name]
else:
b_name = ori_b_name
b_name = tool.brand_clean(b_name)
ext_band_name = self.english_brand_extension(b_name)
for s_name in special_brand_list:
s_name_list = s_name.strip().split("/")
for s_name_item in s_name_list:
idx += 1
if idx % 1000000 == 0: print("idx: %s" % idx)
if len(ext_band_name.lower().split(
s_name_item)) > 1:
# 单个“后”字召回的品牌错误率很高
if s_name_item == "后":
continue
k = "\t".join([
b_id, ori_b_name, ext_band_name, cat1_id,
cat1, gmv
])
recall_brand_dict[k] = ''
r_lst = list(recall_brand_dict.keys()) + mijia_lst
with open(output_file, "w", encoding="utf-8") as f1:
f1.write("\n".join(r_lst))
f1.flush()
except Exception as e:
raise e
def brand_recall(self, output_file="brand_recall_info_tmp_.txt"):
try:
ex_brand_dict = tool.get_exchange_brand_pair()
special_brand_list = self.special_brand_list
recall_brand_dict = {}
mijia_lst = []
pr_list = []
luolamima_list = []
boshikou_list = []
idx = 0
with open(self.standard_brand_file, "r", encoding="utf-8") as f1:
for line in f1:
line = line.strip()
if line == "": continue
lst1 = line.split("\t")
if len(lst1) != 5:
continue
lst1 = [tmp.strip() for tmp in lst1]
b_id, ori_b_name, cat1_id, cat1, gmv = lst1
if self.del_brand_func(b_id): continue
# 米家扩展
mijia_str = self.mijia_special_brand_recall(
b_id, cat1_id, cat1)
if mijia_str != "": mijia_lst.append(mijia_str)
# 红米扩展
redmi_str = self.redmi_special_brand_recall(
b_id, cat1_id, cat1)
if redmi_str != "": mijia_lst.append(redmi_str)
if ori_b_name in ex_brand_dict:
b_name = ex_brand_dict[ori_b_name]
else:
b_name = ori_b_name
b_name = tool.brand_clean(b_name)
ext_band_name = self.english_brand_extension(b_name)
for s_name in special_brand_list:
s_name_list = s_name.strip().split("/")
for s_name_item in s_name_list:
s_name_item = s_name_item.strip()
if s_name_item == "": continue
idx += 1
if idx % 1000000 == 0: print("idx: %s" % idx)
ext_band_name = ext_band_name.lower()
if len(ext_band_name.split(s_name_item)) > 1:
# 单个“后”字召回的品牌错误率很高
if s_name_item == "后":
continue
if b_id == '10414486' and cat1_id == '100010':
ok = 1
'''
99714433 蓝月亮（纸品） 100012 家庭清洁 / 纸品 1685455.83320000
'''
if b_id == '99714433': continue
if not self.brand_cat1_relation_del(
b_id, cat1_id):
k = "\t".join([
b_id, ori_b_name, ext_band_name,
cat1_id, cat1, gmv
])
recall_brand_dict[k] = ''
# 强制添加二手商品的一级类目
if not self.brand_cat1_relation_del(
b_id, "100040"):
k_tmp = "\t".join([
b_id, ori_b_name, ext_band_name,
"100040", "二手商品", '0.0'
])
recall_brand_dict[k_tmp] = ''
# 硬添加【米家】添加一级类目【家用电器】
mijia_lst += [
"\t".join([
"10698337", "MJ/米家",
self.english_brand_extension("MJ/米家/小米米家"), "100031",
"家用电器", "0.0"
])
]
mijia_lst += [
"\t".join([
"10698337", "MJ/米家",
self.english_brand_extension("MJ/米家/小米米家"), "100034",
"家装建材", "0.0"
])
]
# 硬添加【PLUS RAPIDE】添加一级类目【服饰内衣,运动户外,鞋靴】
pr_list += [
"\t".join([
"7830374549", "PLUS RAPIDE",
self.english_brand_extension(
"PLUS RAPIDE/GXG旗下品牌PR/GXG旗下PR/GXG旗下男装PR"), "100006",
"服饰内衣", "0.0"
])
]
pr_list += [
"\t".join([
"7830374549", "PLUS RAPIDE",
self.english_brand_extension(
"PLUS RAPIDE/GXG旗下品牌PR/GXG旗下PR/GXG旗下男装PR"), "100008",
"运动户外", "0.0"
])
]
pr_list += [
"\t".join([
"7830374549", "PLUS RAPIDE",
self.english_brand_extension(
"PLUS RAPIDE/GXG旗下品牌PR/GXG旗下PR/GXG旗下男装PR"), "100001",
"鞋靴", "0.0"
])
]
# 硬添加【罗拉密码】添加一级类目【服饰内衣、鞋靴】
luolamima_list += [
"\t".join([
"99714457", "LOORA PWD/罗拉密码",
self.english_brand_extension("LOORA PWD/罗拉密码"), "100006",
"服饰内衣", "0.0"
])
]
luolamima_list += [
"\t".join([
"99714457", "LOORA PWD/罗拉密码",
self.english_brand_extension("LOORA PWD/罗拉密码"), "100001",
"鞋靴", "0.0"
])
]
# 硬添加【ROYALAPOTHIC/泊诗蔻】添加一级类目【美妆护肤】
boshikou_list += [
"\t".join([
"11015914", "ROYALAPOTHIC/泊诗蔻",
self.english_brand_extension("ROYALAPOTHIC/泊诗蔻"), "100007",
"美妆护肤", "0.0"
])
]
# 品牌丢失一级类目的情况
miss_lst = self.brand_miss_cat1()
r_lst = list(
recall_brand_dict.keys()
) + mijia_lst + miss_lst + pr_list + luolamima_list + boshikou_list
with open(output_file, "w", encoding="utf-8") as f1:
f1.write("\n".join(r_lst))
f1.flush()
except Exception as e:
raise e