def __init__(self):
super(Title_filter, self).__init__()
self.us_btgs_api = US_btgs()
self.amz_attrs_api = AMZ_attrs()
self.amz_sub_attrs_api = AMZ_sub_attrs()
self.amz_title_dic = AMZ_title_dic()
self.titles = Products()
def get_title_dic_by_attr_id(self, attr_id):
query = {}
title_dic = []
query['attr_id'] = attr_id
api_instance = AMZ_title_dic()
try: # get sub_attr_id
r = self.amz_attrs.find_one(query)
for sub_attr_id in r['sub_attr_ids']:
try: # get title_dic by sub_attr_id
offset = 0
limit = 50
while True:
res = api_instance.get_words_by_sub_attr_id(
sub_attr_id, offset=offset, limit=limit)
if limit > len(res):
break
else:
offset = offset + limit
title_dic.extend(res)
except Exception as e:
print(e)
except Exception as e:
print(e)
return title_dic
def get_dictionary_sub_attrs_words(subAttrIds):
res = GetDictionarySubAttrsWordsResponse()
try:
title_dic_api = AMZ_title_dic()
sub_attr_list = []
for subAttrId in subAttrIds:
title_dic_res = title_dic_api.get_dic_by_sub_attr_id(
subAttrId, offset=0, limit=1000)
if title_dic_res:
for a_title_dic in title_dic_res:
title_dic_instance = TitleDic()
title_dic_instance.sub_attr_id = a_title_dic.get(
'sub_attr_id')
title_dic_instance.dic_word = a_title_dic.get(
'sub_attr_dic_word')
title_dic_instance.count = a_title_dic.get('count')
sub_attr_list.append(title_dic_instance)
res.data = sub_attr_list
res.message = 'Successful'
response_status = 200
except Exception as e:
res.message = str(e)
response_status = 400
return res, response_status
def get_dictionary_products_attrs_sub_attrs_words(nodeId, attrId,
subAttrId, word):
res = GetDictionaryProductsAttrsSubAttrsWordsResponse()
try:
title_dic_api = AMZ_title_dic()
title_dic_api_res = title_dic_api.get_dic_by_sub_attr_id(
sub_attr_id=subAttrId, offset=0, limit=1000)
if title_dic_api_res:
title_dic_instance = TitleDic()
is_exist = False
for a_titls_dic in title_dic_api_res:
if a_titls_dic.get('sub_attr_dic_word') == word:
title_dic_instance.sub_attr_id = a_titls_dic.get(
'sub_attr_id')
title_dic_instance.dic_word = a_titls_dic.get(
'sub_attr_dic_word')
title_dic_instance.count = a_titls_dic.get('count')
is_exist = True
if is_exist:
res.data = title_dic_instance
res.message = 'Word already enrolled'
else:
res.message = 'Available'
else:
res.message = 'SubAttrId not exists'
response_status = 200
except Exception as e:
res.message = str(e)
response_status = 400
return res, response_status
def post_dictionary_sub_attrs_words_count_reset(subAttrIds):
res = PostDictionarySubAttrsWordsCountResetResponse()
try:
title_dic_api = AMZ_title_dic()
num = 0
for sub_attr_id in subAttrIds:
title_dic_api_res = title_dic_api.reset_count_to_zero_by_sub_attr_id(
sub_attr_id=sub_attr_id)
num += 1
if len(subAttrIds) == num:
res.message = 'Successful'
response_status = 200
else:
res.message = 'Success partially'
response_status = 200
except Exception as e:
res.message = str(e)
response_status = 400
return res, response_status
from __future__ import print_function
from bl_product_amaz.amz_title_dic import AMZ_title_dic
from pprint import pprint
api_instance = AMZ_title_dic()
try:
offset = 0
limit = 50
sub_attr_id = "aa0000003"
while True:
res = api_instance.get_words_by_sub_attr_id(sub_attr_id, offset, limit)
if limit > len(res):
break
else:
offset = offset + limit
pprint(res)
except Exception as e:
print("Exception when calling get_words_bt_sub_attr_id %s\n" % e)
class Title_filter(DataBase):
def __init__(self):
super(Title_filter, self).__init__()
self.us_btgs_api = US_btgs()
self.amz_attrs_api = AMZ_attrs()
self.amz_sub_attrs_api = AMZ_sub_attrs()
self.amz_title_dic = AMZ_title_dic()
self.titles = Products()
def get_title_word_dic_by_node_id(self, node_ids):
amz_attr_list = []
filter_list = []
# get sub_attrs text and make filter
for node_id in node_ids:
res = self.us_btgs_api.get_btg_by_node_id(node_id)
for amz_attr in res['attr_ids']:
amz_attr_list.append(amz_attr)
#reset_count_zero
for amz_attr_id in amz_attr_list:
res2 = self.amz_attrs_api.get_attr_by_attr_id(amz_attr_id)
sub_attrs = res2['sub_attr_ids']
for sub_attr_id in sub_attrs:
self.amz_title_dic.reset_count_to_zero_by_sub_attr_id(
sub_attr_id)
for amz_attr_id in amz_attr_list:
res1 = self.amz_attrs_api.get_title_dic_by_attr_id(amz_attr_id)
filter_list.extend(res1)
filter_list = sorted(filter_list, key=len)
for i in range(len(filter_list)):
filter_list[i] = filter_list[i].lower()
return filter_list
def filtering_titles(self, node_ids, filter_list):
filtered_titles = []
titles = []
offset = 0
limit = 100
# get titles
for node_id in node_ids:
try:
r = self.titles.get_products_by_node_id(node_id, offset, limit)
for product in r:
title_dic = {}
title_dic['ASIN'] = product['ASIN']
title_dic['brand'] = product['Brand']
title_dic['title'] = product['Title']
titles.append(title_dic)
except Exception as e:
print(e)
count_dic_list = []
# filtering by title dic
for title_info in titles:
title = title_info['title']
# remove brand
title = title.replace(title_info['brand'], "")
title = title.lower()
title = title.replace("(", "")
title = title.replace(")", "")
title = title.replace(",", " ")
for filter in filter_list: # 길이 순서대로
count = 0
len_title = len(title)
title = re.sub('\\b' + filter + ' ' + '\\b', "", title)
len_filter = len(filter) + 1
len_filtered_title = len(title)
if len_title != len_filtered_title:
count_dic = {}
sum_count = count + int(
(len_title - len_filtered_title) / len_filter)
count_dic['word'] = filter
count_dic['count'] = sum_count
count_dic_list.append(count_dic)
title_info['filtered_title'] = title
for tmp in count_dic_list:
self.amz_title_dic.add_count_by_sub_attr_dic_word(
tmp['word'], tmp['count'])
# data clouding
for title in titles:
tmp = title['filtered_title'].split(" ")
filtered_titles.extend(tmp)
result_words = Counter(filtered_titles)
return titles, result_words
def add_sub_attr_in_amz_sub_attrs(self, node_id, attr_id, attr_kr_name,
attr_us_name, sub_attr_id,
sub_attr_kr_name, sub_attr_us_name):
# check attr_id is in amz_attrs DB
# add attr_id in amz_attrs DB and us_btg that node_id is input node_id
self.amz_attrs_api.add_attr(attr_id, attr_kr_name, attr_us_name)
self.us_btgs_api.update_attr_id_by_node_id(node_id, attr_id)
if sub_attr_id != None:
# add sub_attr in amz_sub_attrs DB
self.amz_sub_attrs_api.add_sub_attr(sub_attr_id, sub_attr_kr_name,
sub_attr_us_name)
self.amz_attrs_api.update_sub_attr_ids(attr_id, sub_attr_id)
def add_sub_attr_word_in_amz_title_dic(self, sub_attr_id, sub_attr_word):
self.amz_title_dic.add_title_dic_word(sub_attr_id, sub_attr_word)
from __future__ import print_function
from bl_product_amaz.amz_title_dic import AMZ_title_dic
from pprint import pprint
api_instance = AMZ_title_dic()
try:
sub_attr_id = "aa0000000"
sub_attr_dic_word = "test"
res = api_instance.add_title_dic_word(sub_attr_id, sub_attr_dic_word)
pprint(res)
except Exception as e:
print("Exception when calling add_title_dic_word %s\n" % e)
from __future__ import print_function
from bl_product_amaz.amz_title_dic import AMZ_title_dic
from pprint import pprint
api_instance = AMZ_title_dic()
try:
sub_attr_id = "aa0000003"
res = api_instance.reset_count_to_zero_by_sub_attr_id(sub_attr_id)
pprint(res)
except Exception as e:
print("Exception when calling add_count_by_sub_attr_code: %s\n" % e)
from __future__ import print_function
from bl_product_amaz.amz_title_dic import AMZ_title_dic
from pprint import pprint
api_instance = AMZ_title_dic()
try:
sub_attr_id = "3/4 sleeve"
res = api_instance.add_count_by_sub_attr_dic_word(sub_attr_id, count_up_num=3)
pprint(res)
except Exception as e:
print("Exception when calling add_count_by_sub_attr_code: %s\n" % e)
def get_dictionary_browse_nodes_all(nodeId):
res = GetDictionaryBrowseNodesAllResponse()
try:
attr_data_list = []
us_btgs_api = US_btgs()
attrs_api = AMZ_attrs()
sub_attrs_api = AMZ_sub_attrs()
title_dic_api = AMZ_title_dic()
us_btgs_res = us_btgs_api.get_attrs_by_node_id(nodeId)
if us_btgs_res:
for anAttr in us_btgs_res:
for key in anAttr.keys():
attrs_res = attrs_api.get_attr_by_attr_id(key)
if attrs_res:
res_data = GetDictionaryBrowseNodesAllResponseData(
)
attr_instance = Attr()
attr_instance.attr_id = attrs_res.get('attr_id')
attr_instance.attr_us_name = attrs_res.get(
'attr_us_name')
attr_instance.attr_kr_name = attrs_res.get(
'attr_kr_name')
sub_attr_ids = attrs_res.get('sub_attr_ids')
if sub_attr_ids:
sub_attr_list = []
for sub_attr_id in sub_attr_ids:
sub_attrs_res = sub_attrs_api.get_sub_attr_by_sub_attr_id(
sub_attr_id)
if sub_attrs_res:
sub_attr = GetDictionaryBrowseNodesAllResponseSubAttrs(
)
sub_attr_instance = SubAttr()
sub_attr_instance.sub_attr_id = sub_attrs_res.get(
'sub_attr_id')
sub_attr_instance.sub_attr_us_name = sub_attrs_res.get(
'sub_attr_us_name')
sub_attr_instance.sub_attr_kr_name = sub_attrs_res.get(
'sub_attr_kr_name')
sub_attr.sub_attr = sub_attr_instance
title_dic_res = title_dic_api.get_dic_by_sub_attr_id(
sub_attr_instance.sub_attr_id,
offset=0,
limit=1000)
if title_dic_res:
sub_attr.title_dics = []
for a_title_dic in title_dic_res:
title_dic_instance = TitleDic()
title_dic_instance.sub_attr_id = a_title_dic.get(
'sub_attr_id')
title_dic_instance.dic_word = a_title_dic.get(
'sub_attr_dic_word')
title_dic_instance.count = a_title_dic.get(
'count')
sub_attr.title_dics.append(
title_dic_instance)
sub_attr_list.append(sub_attr)
res_data.attr = attr_instance
res_data.sub_attrs = sub_attr_list
attr_data_list.append(res_data)
res.message = 'Successful'
response_status = 200
res_data = attr_data_list
res.data = res_data
else:
res.message = 'No attributes for node_id: ' + nodeId
response_status = 400
except Exception as e:
res.message = str(e)
response_status = 400
return res, response_status