def brand_reg(self, s_name, brand_lst, brand_id):
r_lst = []
for ext_bname in brand_lst:
if tool.is_all_eng(ext_bname) and tool.is_own_eng(s_name) and ext_bname in s_name:
en_reg_bname = self.english_brand_recognition(ext_bname, s_name)
if en_reg_bname != None:
reg_bname = en_reg_bname
else:
continue
elif ext_bname in s_name:
reg_bname = ext_bname
else:
continue
r_lst.append((reg_bname, brand_id))
#rule_opt_lst = self.rule_opt(s_name, r_lst)
#return rule_opt_lst
return r_lst
def brand_recognition(self, line_str):
'''
brand_recognition is updated by gcw in 2020.09.12.
'''
try:
line = line_str.strip()
if line == "": return None, None, None, None, None, None, None
lst_z = line.split("\001")
if len(lst_z) != 4: return None, None, None, None, None, None, None
lst1 = [tmp.strip() for tmp in lst_z]
product_id, ori_product_name, cat2_id, cat2_name = lst1
# **import
cat1_id, cat1_name = cat2_id, cat2_name
# s_name小写字母
s_name = tool.s_name_dealing(ori_product_name)
if self.brand_rule_obj.no_brand_word_func(s_name):
return None, None, None, None, None, None, None
# 删除
s_name = self.brand_rule_obj.product_name_del_word_func(s_name)
fixed_flag, r_bid = self.brand_rule_obj.fixed_point_func(
product_id)
if fixed_flag:
if r_bid == None:
return None, None, None, None, None, None, None
else:
brand_cat1_id, brand_cat1_name = self.getting_cat1_info(
r_bid)
return r_bid, self.name_ori_brand_dict[r_bid], "", \
brand_cat1_id, brand_cat1_name, cat1_id, cat1_name
else:
pass
clean_brand_2_idx = {}
clean_brand_lst = []
# 1、清洗后的品牌
# 2、清洗后的品牌对应的多个品牌编号
# 1、清洗后的品牌
# 2、清洗后的品牌对应的多个品牌编号
for ori_b_str, same_brand_id_lst in self.brand_idx_dict.items():
# 品牌小写字母
b_str = ori_b_str.lower()
lst_z = b_str.split('|')
if len(lst_z) != 2: continue
b, is_eng = lst_z
if is_eng == "0" and tool.is_own_eng(s_name) and b in s_name:
tmp_b = self.english_brand_recognition(b, s_name)
if tmp_b != None:
for tmp_bid in same_brand_id_lst:
clean_brand_lst.append(b_str)
clean_brand_2_idx[b_str] = tmp_bid
elif b in s_name:
for tmp_bid in same_brand_id_lst:
clean_brand_lst.append(b_str)
clean_brand_2_idx[b_str] = tmp_bid
else:
continue
# 去除已识别品牌中的包含关系
clean_brand_lst = self.brand_inclusion_relation_dealing(
clean_brand_lst)
clean_idx_2_brand = {}
for xxx in clean_brand_lst:
x_id = clean_brand_2_idx[xxx]
if x_id in clean_brand_2_idx:
zzz = clean_idx_2_brand[x_id]
clean_idx_2_brand[x_id] = zzz + [xxx]
else:
clean_idx_2_brand[x_id] = [xxx]
if len(clean_brand_lst) == 0:
return None, None, "没有匹配到标准品牌", None, None, None, None
# 人为规则的过滤
clean_brand_lst, clean_brand_id_lst = self.rule_opt(clean_brand_lst, clean_idx_2_brand, \
ori_product_name, cat1_id, cat1_name)
if len(clean_brand_lst) == 0:
return None, None, "没有匹配到标准品牌", None, None, None, None
if len(clean_brand_id_lst) == 1:
pre_brand_id = clean_brand_id_lst[0]
pre_brand = self.name_ori_brand_dict[pre_brand_id]
match_type = "匹配到唯一标准品牌"
else:
len_brand_dict = {}
for tmp in clean_brand_lst:
l = len(tmp)
if l in len_brand_dict:
x = len_brand_dict[l]
len_brand_dict[l] = x + [tmp]
else:
len_brand_dict[l] = [tmp]
r_lst = [(k, v) for k, v in len_brand_dict.items()]
r_lst = sorted(r_lst, key=lambda m: m[0], reverse=True)
for u in r_lst:
tmp_l, tmp_clean_brand_lst = u
pre_brand_id, pre_brand, match_type = self.same_cat1_strategy(
cat1_id, tmp_clean_brand_lst, clean_brand_id_lst)
if pre_brand == None or pre_brand_id == None: #上方的选择一级类下高GMV品牌策略失效，没有匹配到相同一级类的品牌，则走如下逻辑
pre_brand_id, pre_brand, match_type = self.same_length_strategy(
tmp_clean_brand_lst, clean_brand_id_lst)
if pre_brand != None and match_type != None:
break
if pre_brand_id != None:
brand_cat1_id, brand_cat1_name = self.getting_cat1_info(
pre_brand_id)
else:
brand_cat1_id, brand_cat1_name = None, None
return pre_brand_id, pre_brand, match_type, \
brand_cat1_id, brand_cat1_name, cat1_id, cat1_name
except Exception as e:
raise e
def brand_same_cat1_recognition(self, line_str):
'''
brand_recognition is updated by gcw in 2020.09.12.
'''
try:
line = line_str.strip()
if line == "": return None, None, None, None, None, None, None
lst_z = line.split("\001")
if len(lst_z) != 5: return None, None, None, None, None, None, None
lst1 = [tmp.strip() for tmp in lst_z]
product_id, ori_product_name, ori_brand_word, cat1_id, cat1_name = lst1
# s_name小写字母
s_name = tool.s_name_dealing("%s %s" % (ori_product_name, ori_brand_word))
s_name = self.product_name_del_word(s_name)
clean_brand_lst = []
clean_brand_id_lst = []
# 1、清洗后的品牌
# 2、清洗后的品牌对应的多个品牌编号
for ori_b_str, same_brand_id_lst in self.brand_idx_dict.items():
# 品牌小写字母
b_str = ori_b_str.lower()
lst_z = b_str.split('|')
if len(lst_z) != 2: continue
b, is_eng = lst_z
if is_eng == "0" and tool.is_own_eng(s_name) and b in s_name:
tmp_b = self.english_brand_recognition(b, s_name)
if tmp_b != None:
clean_brand_lst.append(ori_b_str)
clean_brand_id_lst += same_brand_id_lst
elif b in s_name:
clean_brand_lst.append(ori_b_str)
clean_brand_id_lst += same_brand_id_lst
else:
continue
clean_brand_lst = list(set(clean_brand_lst))
clean_brand_id_lst = list(set(clean_brand_id_lst))
if len(clean_brand_lst) == 0: return None, None, "没有匹配到标准品牌", None, None, None, None
# 人为规则的过滤
clean_brand_lst, clean_brand_id_lst = self.rule_opt(clean_brand_lst, ori_product_name, cat1_id, cat1_name)
if len(clean_brand_lst) == 0:
return None, None, "没有匹配到标准品牌", None, None, None, None
if len(clean_brand_id_lst) == 1:
b_id = clean_brand_id_lst[0]
brand_cat1_id, brand_cat1_name = self.getting_cat1_info(b_id)
return b_id, \
self.name_ori_brand_dict[clean_brand_id_lst[0]], \
"匹配到唯一标准品牌", brand_cat1_id, brand_cat1_name, cat1_id, cat1_name
pre_brand_id, pre_brand, match_type = None, None, None
if len(clean_brand_lst) == 1:
pre_brand_id, pre_brand, match_type = \
self.same_cat1_gmv_strategy(cat1_id, clean_brand_lst, clean_brand_id_lst)
else:
len_brand_dict = {}
for tmp in clean_brand_lst:
l = len(tmp)
if l in len_brand_dict:
x = len_brand_dict[l]
len_brand_dict[l] = x + [tmp]
else:
len_brand_dict[l] = [tmp]
r_lst = [(k, v) for k, v in len_brand_dict.items()]
r_lst = sorted(r_lst, key=lambda m: m[0], reverse=True)
for u in r_lst:
tmp_l, tmp_clean_brand_lst = u
pre_brand_id, pre_brand, match_type = self.same_cat1_strategy(cat1_id, tmp_clean_brand_lst,
clean_brand_id_lst)
if pre_brand != None and match_type != None:
break
# 兜底处理
if pre_brand == None and match_type == None:
pre_brand_id, pre_brand, match_type = \
self.same_cat1_gmv_strategy(cat1_id, clean_brand_lst, clean_brand_id_lst)
if pre_brand_id != None:
brand_cat1_id, brand_cat1_name = self.getting_cat1_info(pre_brand_id)
else:
brand_cat1_id = None
brand_cat1_name = None
return pre_brand_id, pre_brand, match_type, \
brand_cat1_id, brand_cat1_name, cat1_id, cat1_name
except Exception as e:
raise e
def brand_reg_opt(self, reg_brand_dict, line_str):
try:
line = line_str.strip()
if line == "": return None, None, None, None, None, None, None
lst_z = line.split("\001")
if len(lst_z) != 5: return None, None, None, None, None, None, None
lst1 = [tmp.strip() for tmp in lst_z]
product_id, ori_product_name, ori_brand_word, cat1_id, cat1_name = lst1
#
# s_name小写字母
s_name = tool.s_name_dealing("%s %s" %
(ori_product_name, ori_brand_word))
s_name = self.product_name_del_word(s_name)
clean_idx_2_brand = {}
# 英语手机品牌清洗，比如：vivo，oppo等的清洗
englisg_phone_brand_lst = self.brand_rule_obj.english_phone_rule_func(
s_name, cat1_id)
for epb in englisg_phone_brand_lst:
epb_id, epb_name = epb
epb_name = epb_name + '|0'
if epb_name not in self.brand_idx_dict: continue
clean_idx_2_brand[epb_id] = epb_name
# 1、清洗后的品牌
# 2、清洗后的品牌对应的多个品牌编号
for ori_b_str, same_brand_id_lst in reg_brand_dict.items():
# 品牌小写字母
b_str = ori_b_str.lower()
lst_z = b_str.split('|')
if len(lst_z) != 2: continue
b, is_eng = lst_z
if is_eng == "0" and tool.is_own_eng(s_name) and b in s_name:
tmp_b = self.english_brand_recognition(b, s_name)
if tmp_b != None:
for tmp_bid in same_brand_id_lst:
clean_idx_2_brand[tmp_bid] = ori_b_str
elif b in s_name:
for tmp_bid in same_brand_id_lst:
clean_idx_2_brand[tmp_bid] = ori_b_str
else:
continue
clean_brand_lst = list(set(list(clean_idx_2_brand.values())))
clean_brand_id_lst = list(set(list(clean_idx_2_brand.keys())))
if len(clean_brand_lst) == 0:
return None, None, "没有匹配到标准品牌", None, None, None, None
# 人为规则的过滤
clean_brand_lst, clean_brand_id_lst = self.rule_opt(
clean_brand_id_lst, clean_idx_2_brand, ori_product_name,
cat1_id, cat1_name)
if len(clean_brand_lst) == 0:
return None, None, "没有匹配到标准品牌", None, None, None, None
if len(clean_brand_id_lst) == 1:
b_id = clean_brand_id_lst[0]
brand_cat1_id, brand_cat1_name = self.getting_cat1_info(b_id)
return b_id, \
self.name_ori_brand_dict[clean_brand_id_lst[0]], \
"匹配到唯一标准品牌", brand_cat1_id, brand_cat1_name, cat1_id, cat1_name
pre_brand_id, pre_brand, match_type = None, None, None
if len(clean_brand_lst) == 1:
pre_brand_id, pre_brand, match_type = \
self.same_cat1_gmv_strategy(cat1_id, clean_brand_lst, clean_brand_id_lst)
else:
len_brand_dict = {}
for tmp in clean_brand_lst:
l = len(tmp)
if l in len_brand_dict:
x = len_brand_dict[l]
len_brand_dict[l] = x + [tmp]
else:
len_brand_dict[l] = [tmp]
r_lst = [(k, v) for k, v in len_brand_dict.items()]
r_lst = sorted(r_lst, key=lambda m: m[0], reverse=True)
for u in r_lst:
tmp_l, tmp_clean_brand_lst = u
pre_brand_id, pre_brand, match_type = self.same_cat1_strategy(
cat1_id, tmp_clean_brand_lst, clean_brand_id_lst)
if pre_brand == None or pre_brand_id == None:
pre_brand_id, pre_brand, match_type = self.same_length_strategy(
tmp_clean_brand_lst, clean_brand_id_lst)
if pre_brand != None and match_type != None:
break
if pre_brand_id != None:
brand_cat1_id, brand_cat1_name = self.getting_cat1_info(
pre_brand_id)
else:
brand_cat1_id = None
brand_cat1_name = None
return pre_brand_id, pre_brand, match_type, \
brand_cat1_id, brand_cat1_name, cat1_id, cat1_name
except Exception as e:
raise e
def brand_recognition(self, line_str):
'''
brand_recognition is updated by gcw in 2020.09.12.
'''
try:
line = line_str.strip()
if line == "": return None, None, None, None, None, None, None
lst_z = line.split("\001")
if len(lst_z) != 5: return None, None, None, None, None, None, None
lst1 = [tmp.strip() for tmp in lst_z]
product_id, ori_product_name, _, cat1_id, cat1_name = lst1
# s_name小写字母
s_name = tool.s_name_dealing(ori_product_name)
s_name = self.product_name_del_word(s_name)
clean_brand_2_idx = {}
clean_brand_lst = []
# 英语手机品牌清洗，比如：vivo，oppo等的清洗
englisg_phone_brand_lst = self.brand_rule_obj.english_phone_rule_func(
s_name, cat1_id)
for epb in englisg_phone_brand_lst:
epb_id, epb_name = epb
epb_name = epb_name + '|0'
if epb_name not in self.brand_idx_dict: continue
clean_brand_2_idx[epb_name] = epb_id
clean_brand_lst.append(epb_name)
# 1、清洗后的品牌
# 2、清洗后的品牌对应的多个品牌编号
for ori_b_str, same_brand_id_lst in self.brand_idx_dict.items():
# 品牌小写字母
b_str = ori_b_str.lower()
lst_z = b_str.split('|')
if len(lst_z) != 2: continue
b, is_eng = lst_z
if is_eng == "0" and tool.is_own_eng(s_name) and b in s_name:
tmp_b = self.english_brand_recognition(b, s_name)
if tmp_b != None:
for tmp_bid in same_brand_id_lst:
clean_brand_2_idx[ori_b_str] = tmp_bid
clean_brand_lst.append(ori_b_str)
elif b in s_name:
for tmp_bid in same_brand_id_lst:
clean_brand_2_idx[ori_b_str] = tmp_bid
clean_brand_lst.append(ori_b_str)
else:
continue
# 去除已识别品牌中的包含关系
clean_brand_lst = self.brand_inclusion_relation_dealing(
clean_brand_lst)
clean_idx_2_brand = {}
for xxx in clean_brand_lst:
x_id = clean_brand_2_idx[xxx]
if x_id in clean_brand_2_idx:
zzz = clean_idx_2_brand[x_id]
clean_idx_2_brand[x_id] = zzz + [xxx]
else:
clean_idx_2_brand[x_id] = [xxx]
if len(clean_brand_lst) == 0:
return None, None, "没有匹配到标准品牌", None, None, None, None
# 人为规则的过滤
clean_brand_lst, clean_brand_id_lst = self.rule_opt(clean_brand_lst, clean_idx_2_brand, \
ori_product_name, cat1_id, cat1_name)
if len(clean_brand_lst) == 0:
return None, None, "没有匹配到标准品牌", None, None, None, None
if len(clean_brand_id_lst) == 1:
b_id = clean_brand_id_lst[0]
brand_cat1_id, brand_cat1_name = self.getting_cat1_info(b_id)
return b_id, \
self.name_ori_brand_dict[clean_brand_id_lst[0]], \
"匹配到唯一标准品牌", brand_cat1_id, brand_cat1_name, cat1_id, cat1_name
pre_brand_id, pre_brand, match_type = None, None, None
if len(clean_brand_lst) == 1:
pre_brand_id, pre_brand, match_type = \
self.same_cat1_gmv_strategy(cat1_id, clean_brand_lst, clean_brand_id_lst)
else:
len_brand_dict = {}
for tmp in clean_brand_lst:
l = len(tmp)
if l in len_brand_dict:
x = len_brand_dict[l]
len_brand_dict[l] = x + [tmp]
else:
len_brand_dict[l] = [tmp]
r_lst = [(k, v) for k, v in len_brand_dict.items()]
r_lst = sorted(r_lst, key=lambda m: m[0], reverse=True)
for u in r_lst:
tmp_l, tmp_clean_brand_lst = u
pre_brand_id, pre_brand, match_type = self.same_cat1_strategy(
cat1_id, tmp_clean_brand_lst, clean_brand_id_lst)
if pre_brand == None or pre_brand_id == None:
pre_brand_id, pre_brand, match_type = self.same_length_strategy(
tmp_clean_brand_lst, clean_brand_id_lst)
if pre_brand != None and match_type != None:
break
if pre_brand_id != None:
brand_cat1_id, brand_cat1_name = self.getting_cat1_info(
pre_brand_id)
else:
brand_cat1_id = None
brand_cat1_name = None
return pre_brand_id, pre_brand, match_type, \
brand_cat1_id, brand_cat1_name, cat1_id, cat1_name
except Exception as e:
raise e