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 english_brand_extension(self, brand_name):
"""
target: 将扩展的品牌直接保存值召回品牌中
1）指定品牌
2）标准品牌
第一种情况：去特殊字符
A.H.C/爱和纯 -> AHC爱和纯 -> A.H.C/爱和纯/AHC爱和纯
A.O.史密斯 -> AO史密斯 -> A.O.史密斯/AO史密斯
第二种情况：去英文的空格
MAKE UP FOR EVER -> MAKEUPFOREVER
COLOR KEY -> COLORKEY
a b c/某某某 -> abc/a b c/某某某/abc某某某
:return:
"""
def _single_brand_ext(tmp_b_name):
# 去除空格
b1 = re.sub(r"[\s]+", "", tmp_b_name)
# 去除.
b2 = tmp_b_name.replace(".", "").replace("．", "")
r_lst = list(set([tmp_b_name, b1, b2]))
return r_lst
# 10943455 Hisense/海信（黑电）
ok_brand_name = ""
tmp = brand_name.strip().replace("（",
"(").replace("）",
"").replace(")", "")
lst2 = tmp.split("(")
if len(lst2) == 2:
b1 = lst2[0]
if tool.is_all_eng(lst2[1]):
b2 = lst2[1]
ok_brand_name = b2 + "/" + b1
else:
ok_brand_name = b1
else:
ok_brand_name = brand_name
brand_lst = ok_brand_name.strip().split("/")
re_brand_lst = []
if len(brand_lst) == 1:
re_brand_lst += _single_brand_ext(brand_lst[0])
else:
en_brand_lst = []
ch_brand_lst = []
other_brand_lst = []
for b in brand_lst:
if tool.is_all_eng(b):
en_brand_lst.append(b)
elif tool.is_all_chinese(b):
ch_brand_lst.append(b)
else:
other_brand_lst.append(b)
en_brand_ext_lst = []
for z in en_brand_lst:
en_brand_ext_lst += _single_brand_ext(z)
mix_brand_lst = []
for y in en_brand_ext_lst:
for x in ch_brand_lst:
mix_brand_lst.append(y + x)
mix_brand_lst.append(x + y)
if len(en_brand_lst) > 1:
for i in range(len(en_brand_ext_lst)):
for j in range(i + 1, len(en_brand_lst)):
mix_brand_lst.append(en_brand_lst[i] + en_brand_lst[j])
mix_brand_lst.append(en_brand_lst[j] + en_brand_lst[i])
re_brand_lst = mix_brand_lst + en_brand_ext_lst + ch_brand_lst + other_brand_lst
re_brand_lst = list(set(re_brand_lst))
#print(re_brand_lst)
return "/".join(re_brand_lst)
def brand_info_loading(self):
cat1_brand_dict = {} # 一级类下包含哪些品牌
cat1_clean_brand_dict = {}
cat1_dict = {}
brand_cat1_dict = {} # {brand_id: [cat1, cat2]}
brand_idx_dict = {} # {"苹果": [1, 2]}
idx_ori_brand_dict = {} #
name_ori_brand_dict = {} # 品牌名称原始字符串
brand_gmv_dict = {} # 品牌的gmv
idx_brand_lst_dict = {
} # 处理后的品牌，比如：[1 Apple/苹果] -》{'1': ['apple', '苹果']}
idx = 0
with open(self._brand_info_file, "r", encoding="utf-8") as f1:
for line in f1:
line = line.strip()
if line == "": continue
if line.startswith("#"): continue
# brand_id, brand_name, cat1_id, cat1, gmv
lst1 = line.split("\t")
if len(lst1) != 6:
continue
lst1 = [tmp.strip() for tmp in lst1]
b_id, b_name_ori, b_name, cat1_id, cat1, gmv = lst1
# idx-brand
idx_ori_brand_dict[b_id] = b_name
if b_id == '10698337':
name_ori_brand_dict[b_id] = 'Xiaomi/小米'
else:
name_ori_brand_dict[b_id] = b_name_ori
cat1_dict[cat1_id] = cat1
if b_name in self._exchange_brand_dict:
b_name = self._exchange_brand_dict[b_name]
r_brand_set = tool.brand_dealing(b_name)
brand_gmv_dict[b_id] = round(float(gmv), 3)
idx_brand_lst_dict[b_id] = list(r_brand_set)
# brand-idx
for r in r_brand_set:
if len(r) == 1: continue
if tool.is_number(r): continue
is_eng = tool.is_all_eng(r)
if is_eng and len(r) < 3: continue
# 需要删除的品牌
if r in self._del_brand_dict: continue
flag = "0" if is_eng else "1"
r = "%s|%s" % (r, flag)
if r in brand_idx_dict:
z = brand_idx_dict[r]
z = [b_id] + z
z = list(set(z))
brand_idx_dict[r] = z
else:
brand_idx_dict[r] = [b_id]
if cat1_id in cat1_clean_brand_dict:
p = cat1_clean_brand_dict[cat1_id]
cat1_clean_brand_dict[cat1_id] = p + [r]
else:
cat1_clean_brand_dict[cat1_id] = [r]
# mkt2-brand
if cat1 != "NULL":
if b_id in brand_cat1_dict:
xx = brand_cat1_dict[b_id]
brand_cat1_dict[b_id] = xx + [cat1_id]
else:
brand_cat1_dict[b_id] = [cat1_id]
if cat1_id in cat1_brand_dict:
lst_9 = cat1_brand_dict[cat1_id]
cat1_brand_dict[cat1_id] = ["%s|%s" %
(b_id, b_name)] + lst_9
else:
cat1_brand_dict[cat1_id] = ["%s|%s" % (b_id, b_name)]
else:
continue
return brand_idx_dict, idx_ori_brand_dict, name_ori_brand_dict,\
brand_cat1_dict, cat1_brand_dict, \
cat1_clean_brand_dict, brand_gmv_dict, cat1_dict
def _is_all_eng(self, s1):
return tool.is_all_eng(s1)