def childes(
selector: SelectorList,
parent_tag: str,
) -> SelectorList:
if not isinstance(parent_tag, str):
raise TypeError('Given `parent_tag` is not `str` object.')
childes_selector = SelectorList()
iterate_selector_string_template = parent_tag + ' > :nth-child({i})'
i = 1
# starting the iteration
while True:
child = selector.css(iterate_selector_string_template.format(i=i))
if child:
childes_selector.append(child)
i += 1
else:
return childes_selector
def parse_product(self, response):
pic = response.xpath('//ul[@id="J_UlThumb"]/li/a/img/@src').extract()
detail = response.xpath('//div[@class="attributes" and @id="attributes"]').extract()
color_pics = response.xpath('//dd/ul[contains(@class,"tm-clear J_TSaleProp tb-img")]/li')
total_list = response.xpath('//dd/ul[contains(@class,"tm-clear J_TSaleProp")]/li')
json_getter = response.xpath('//div[@class="tm-clear"]/script[3]').extract()
extract_pic =
print len(extract_pic)
size_list = SelectorList()
for i in total_list:
if i.xpath('a/span/text()').extract()[0] not in color_pics.xpath('a/span/text()').extract():
size_list.append(i)
# cnt = 0
# sku_type = 0
# for i in total_list:
# cnt = cnt + 1
# if i.xpath('a/span/text()').extract()[0] not in color_pics.xpath('a/span/text()').extract():
# size_list.append(i)
# sku_type = cnt
# if sku_type > len(total_list)-len(color_pics):
# sku_type = 0
# else:
# sku_type = 1
#
# if len(size_list)==0:
# sku_type = 3
#
# cut_line = len(total_list)-len(color_pics)
str_val_map = {}
for i in range(len(total_list)):
v = total_list[i].xpath('@data-value').extract()
n = total_list[i].xpath('a/span/text()').extract()
if len(n)>0 and len(v)>0:
str_val_map[n[0]] = v[0]
else:
print "Value Error"
#size_list = total_list[:cut_line]
st = json_getter[0].split('TShop.Setup(')[1]
info_dict = self.python_getter(st)
skuMap = info_dict["valItemInfo"]["skuMap"]
product_img = []
for i in pic:
tmp = self.resize_pic(i)
product_img.append(tmp)
# color_set = selen.single_page(response.url)
color_set = []
for i in color_pics:
color = {}
color['color'] = i.xpath('@title').extract()[0]
try:
tmp_str = i.xpath('a/@style').extract()[0]
tmp_str = tmp_str[tmp_str.find('(')+3:tmp_str.rfind(')')]
tmp_str = tmp_str[0:tmp_str.rfind('_')]
color['image_url'] = tmp_str
except:
color['image_url'] = product_img[0]
color['alternative_image_urls'] = product_img
color['pricing_list'] = self.get_pricing(skuMap, str_val_map, i, size_list, response.meta['price'])
if len(color['pricing_list'])>0:
color_set.append(color)
item = MTSGetdataItem()
item['product_url'] = response.meta['product_url']
item['item_id'] = response.meta['item_id']
item['title'] = response.meta['title']
item['brand'] = 'Midi'
item['merchant'] = 'Tmall'
item['product_description'] = ''
item['product_detail'] = detail[0]
item['colors'] = color_set
item['categories'] = info_dict['itemDO']['categoryId']
if len(item['colors'])>0:
yield item