326 lines
16 KiB
Python
326 lines
16 KiB
Python
from core import spiders, types
|
|
from utils import formats, messages
|
|
|
|
class SpiderModule(spiders.Spiders):
|
|
project_name = "ancientnutrition"
|
|
|
|
def __init__(self):
|
|
super().__init__(
|
|
self.project_name,
|
|
bitch = 30,
|
|
worker_num = 1,
|
|
switch_clash = False,
|
|
reflush_browser = False
|
|
)
|
|
|
|
def get_category_urls(self) -> list[str]:
|
|
category_urls = []
|
|
|
|
self.tab.get('https://ancientnutrition.com/')
|
|
one_menu_eles = self.tab.eles('xpath=//*[@id="__next"]/div[4]/div[1]/div[1]/div')
|
|
for index, one_menu_ele in enumerate(one_menu_eles[1:]):
|
|
one_link_ele = one_menu_ele.ele('xpath=div/a')
|
|
one_url = one_link_ele.attr('href')
|
|
one_name = one_link_ele.text.replace('/', '-')
|
|
category_urls.append(f"{one_url}#{one_name}")
|
|
messages.sendInfo(one_name)
|
|
|
|
print(index+2)
|
|
two_menu_eles = self.tab.eles(f'xpath=//*[@id="__next"]/div[4]/div[{index+2}]/div/div/div[2]/div/a')
|
|
for two_menu_ele in two_menu_eles:
|
|
two_link_ele = two_menu_ele
|
|
two_url = two_link_ele.attr('href')
|
|
two_name = two_link_ele.text.replace('/', '-')
|
|
category_urls.append(f"{two_url}#{one_name}/{two_name}")
|
|
|
|
# three_menu_eles = two_menu_ele.eles('xpath=ul/li')
|
|
# for three_menu_ele in three_menu_eles:
|
|
# three_link_ele = three_menu_ele.ele('@tag()=a')
|
|
# three_url = three_link_ele.attr('href')
|
|
# three_name = three_link_ele.text.replace('/', '-')
|
|
# category_urls.append(f"{three_url}#{one_name}/{two_name}/{three_name}")
|
|
|
|
return category_urls
|
|
|
|
def get_goods_urls(self, category_url: str) -> list[str]:
|
|
self.tab.get(f"{category_url}")
|
|
goods_urls = []
|
|
|
|
while True:
|
|
try:
|
|
self.tab.scroll.to_bottom()
|
|
self.tab.ele('text=Load More').click()
|
|
except:
|
|
break
|
|
|
|
goods_eles = self.tab.eles('xpath=//*[@id="__next"]/div[9]/div/div[2]/div[2]/div')
|
|
for goods_ele in goods_eles:
|
|
try:
|
|
goods_link_ele = goods_ele.ele('@tag()=a')
|
|
except Exception as e:
|
|
continue
|
|
goods_urls.append(goods_link_ele.attr('href'))
|
|
|
|
return goods_urls
|
|
|
|
def get_goods_info(self, url) -> types.GoodsInfo:
|
|
def get_image_urls():
|
|
image_urls = []
|
|
|
|
image_eles = tab.eles('xpath=//*[@id="__next"]/div[8]/div[1]/div/div[2]/div[1]/div/div[2]/div/div[1]/div')
|
|
for image_ele in image_eles:
|
|
image_link_ele = image_ele.ele('xpath=div/span/img')
|
|
image_urls.append(image_link_ele.attr('src').replace('_150x150', '_1450x1450'))
|
|
|
|
return image_urls
|
|
|
|
tab = self.browser.new_tab(url)
|
|
tab.set.window.max()
|
|
spu = formats.url_to_spu(url)
|
|
|
|
try:
|
|
brand = ''
|
|
# try:
|
|
# brand = tab.ele('xpath=//*[@id="contentContainer"]/form/div[2]/div[1]/h1/div[1]', timeout=10).text
|
|
# except:
|
|
# pass
|
|
# if brand == '':
|
|
# title = tab.ele('xpath=//*[@id="contentContainer"]/form/div[2]/div[1]/h1').text
|
|
# else:
|
|
title = tab.ele('@tag()=h1').text
|
|
|
|
desc = formats.clean_html(tab.ele('#productDescription').html)
|
|
|
|
price = float(tab.ele('xpath=//*[@id="__next"]/div[8]/div[1]/div/div[2]/div[2]/div/div[1]/div/span/span[1]').text.replace('$', ''))
|
|
old_price = float(tab.ele('xpath=//*[@id="__next"]/div[8]/div[1]/div/div[2]/div[2]/div/div[1]/div/span/span[2]/span[1]').text.replace('$', ''))
|
|
|
|
main_images = get_image_urls()
|
|
|
|
goods_info = types.GoodsInfo(
|
|
title=title,
|
|
desc=desc,
|
|
brand=brand,
|
|
url=url,
|
|
spu=spu,
|
|
price=price,
|
|
old_price=old_price,
|
|
)
|
|
|
|
m_data = []
|
|
attr_items = []
|
|
try:
|
|
tab.ele('text=Select Flavor').click()
|
|
m_data.append('Flavor')
|
|
except:
|
|
pass
|
|
try:
|
|
tab.ele('text=Select Size').click()
|
|
m_data.append('Size')
|
|
except:
|
|
pass
|
|
try:
|
|
tab.ele('text=Select Quantity').click()
|
|
m_data.append('Quantity')
|
|
except:
|
|
pass
|
|
|
|
if 'Flavor' in m_data:
|
|
tab.ele('xpath=//*[@id="__next"]/div[8]/div[1]/div/div[2]/div[2]/div/div[3]/div/div').click()
|
|
color_eles = tab.eles('xpath=//*[@id="__next"]/div[8]/div[1]/div/div[2]/div[2]/div/div[3]/div/div[2]/div')
|
|
color_xpaths = []
|
|
for color_ele in color_eles:
|
|
color_xpaths.append(color_ele.xpath)
|
|
|
|
for color_xpath in color_xpaths:
|
|
while True:
|
|
tab.ele('xpath=//*[@id="__next"]/div[8]/div[1]/div/div[2]/div[2]/div/div[3]/div/div').click()
|
|
try:
|
|
color_ele = tab.ele(f"xpath={color_xpath}")
|
|
color_name = color_ele.text
|
|
break
|
|
except:
|
|
tab.get(url)
|
|
color_ele.click()
|
|
tab.wait(3)
|
|
color_images = get_image_urls()
|
|
if 'Size' in m_data:
|
|
try:
|
|
size_eles = tab.eles(f"xpath={tab.ele('text=Select Size').xpath.replace('/span[1]', '/div/div')}")
|
|
except:
|
|
size_name = 'Default'
|
|
if 'Quantity' in m_data:
|
|
quantity_eles = tab.eles(f"xpath={tab.ele('text=Select Quantity').xpath.replace('/span[1]', '/div/div')}")
|
|
for quantity_index, quantity_ele in enumerate(quantity_eles):
|
|
try:
|
|
quantity_ele = tab.ele(f"xpath={tab.ele('text=Select Quantity').xpath.replace('/span[1]', f'/div/div[{quantity_index+1}]')}")
|
|
quantity_name = quantity_ele.text
|
|
except:
|
|
continue
|
|
quantity_ele.click()
|
|
tab.wait(2)
|
|
price = float(tab.ele('xpath=//*[@id="__next"]/div[8]/div[1]/div/div[2]/div[2]/div/div[1]/div/span/span[1]').text.replace('$', ''))
|
|
old_price = float(tab.ele('xpath=//*[@id="__next"]/div[8]/div[1]/div/div[2]/div[2]/div/div[1]/div/span/span[2]/span[1]').text.replace('$', ''))
|
|
attr_items.append({
|
|
'url': tab.url,
|
|
'items': [color_name, size_name, quantity_name],
|
|
'price': price,
|
|
'old_price': old_price,
|
|
'images': color_images
|
|
})
|
|
else:
|
|
price = float(tab.ele('xpath=//*[@id="__next"]/div[8]/div[1]/div/div[2]/div[2]/div/div[1]/div/span/span[1]').text.replace('$', ''))
|
|
old_price = float(tab.ele('xpath=//*[@id="__next"]/div[8]/div[1]/div/div[2]/div[2]/div/div[1]/div/span/span[2]/span[1]').text.replace('$', ''))
|
|
attr_items.append({
|
|
'url': tab.url,
|
|
'items': [color_name, size_name],
|
|
'price': price,
|
|
'old_price': old_price,
|
|
'images': color_images
|
|
})
|
|
size_xpaths = []
|
|
for size_ele in size_eles:
|
|
try:
|
|
size_xpaths.append(size_ele.xpath)
|
|
except:
|
|
continue
|
|
|
|
for size_xpath in size_xpaths:
|
|
try:
|
|
size_ele = tab.ele(f"xpath={size_xpath}")
|
|
size_name = size_ele.text
|
|
except:
|
|
continue
|
|
size_ele.click()
|
|
tab.wait(3)
|
|
color_images = get_image_urls()
|
|
if 'Quantity' in m_data:
|
|
quantity_eles = tab.eles(f"xpath={tab.ele('text=Select Quantity').xpath.replace('/span[1]', '/div/div')}")
|
|
for quantity_index, quantity_ele in enumerate(quantity_eles):
|
|
try:
|
|
quantity_ele = tab.ele(f"xpath={tab.ele('text=Select Quantity').xpath.replace('/span[1]', f'/div/div[{quantity_index+1}]')}")
|
|
quantity_name = quantity_ele.text
|
|
except:
|
|
continue
|
|
quantity_ele.click()
|
|
tab.wait(2)
|
|
price = float(tab.ele('xpath=//*[@id="__next"]/div[8]/div[1]/div/div[2]/div[2]/div/div[1]/div/span/span[1]').text.replace('$', ''))
|
|
old_price = float(tab.ele('xpath=//*[@id="__next"]/div[8]/div[1]/div/div[2]/div[2]/div/div[1]/div/span/span[2]/span[1]').text.replace('$', ''))
|
|
attr_items.append({
|
|
'url': tab.url,
|
|
'items': [color_name, size_name, quantity_name],
|
|
'price': price,
|
|
'old_price': old_price,
|
|
'images': color_images
|
|
})
|
|
else:
|
|
price = float(tab.ele('xpath=//*[@id="__next"]/div[8]/div[1]/div/div[2]/div[2]/div/div[1]/div/span/span[1]').text.replace('$', ''))
|
|
old_price = float(tab.ele('xpath=//*[@id="__next"]/div[8]/div[1]/div/div[2]/div[2]/div/div[1]/div/span/span[2]/span[1]').text.replace('$', ''))
|
|
attr_items.append({
|
|
'url': tab.url,
|
|
'items': [color_name, size_name],
|
|
'price': price,
|
|
'old_price': old_price,
|
|
'images': color_images
|
|
})
|
|
else:
|
|
price = float(tab.ele('xpath=//*[@id="__next"]/div[8]/div[1]/div/div[2]/div[2]/div/div[1]/div/span/span[1]').text.replace('$', ''))
|
|
old_price = float(tab.ele('xpath=//*[@id="__next"]/div[8]/div[1]/div/div[2]/div[2]/div/div[1]/div/span/span[2]/span[1]').text.replace('$', ''))
|
|
attr_items.append({
|
|
'url': tab.url,
|
|
'items': [color_name],
|
|
'price': price,
|
|
'old_price': old_price,
|
|
'images': color_images
|
|
})
|
|
else:
|
|
color_images = []
|
|
if 'Size' in m_data:
|
|
size_eles = tab.eles(f"xpath={tab.ele('text=Select Size').xpath.replace('/span[1]', '/div/div')}")
|
|
size_xpaths = []
|
|
for size_ele in size_eles:
|
|
try:
|
|
size_xpaths.append(size_ele.xpath)
|
|
except:
|
|
continue
|
|
|
|
for size_xpath in size_xpaths:
|
|
try:
|
|
size_ele = tab.ele(f"xpath={size_xpath}")
|
|
size_name = size_ele.text
|
|
except:
|
|
continue
|
|
size_ele.click()
|
|
tab.wait(2)
|
|
color_images = get_image_urls()
|
|
if 'Quantity' in m_data:
|
|
quantity_eles = tab.eles(f"xpath={tab.ele('text=Select Quantity').xpath.replace('/span[1]', '/div/div')}")
|
|
for quantity_index, quantity_ele in enumerate(quantity_eles):
|
|
try:
|
|
quantity_ele = tab.ele(f"xpath={tab.ele('text=Select Quantity').xpath.replace('/span[1]', f'/div/div[{quantity_index+1}]')}")
|
|
quantity_name = quantity_ele.text
|
|
except:
|
|
continue
|
|
quantity_ele.click()
|
|
tab.wait(2)
|
|
price = float(tab.ele('xpath=//*[@id="__next"]/div[8]/div[1]/div/div[2]/div[2]/div/div[1]/div/span/span[1]').text.replace('$', ''))
|
|
old_price = float(tab.ele('xpath=//*[@id="__next"]/div[8]/div[1]/div/div[2]/div[2]/div/div[1]/div/span/span[2]/span[1]').text.replace('$', ''))
|
|
attr_items.append({
|
|
'url': tab.url,
|
|
'items': [size_name, quantity_name],
|
|
'price': price,
|
|
'old_price': old_price,
|
|
'images': color_images
|
|
})
|
|
else:
|
|
if 'Quantity' in m_data:
|
|
quantity_eles = tab.eles(f"xpath={tab.ele('text=Select Quantity').xpath.replace('/span[1]', '/div/div')}")
|
|
for quantity_index, quantity_ele in enumerate(quantity_eles):
|
|
try:
|
|
quantity_ele = tab.ele(f"xpath={tab.ele('text=Select Quantity').xpath.replace('/span[1]', f'/div/div[{quantity_index+1}]')}")
|
|
quantity_name = quantity_ele.text
|
|
except:
|
|
continue
|
|
quantity_ele.click()
|
|
tab.wait(2)
|
|
price = float(tab.ele('xpath=//*[@id="__next"]/div[8]/div[1]/div/div[2]/div[2]/div/div[1]/div/span/span[1]').text.replace('$', ''))
|
|
old_price = float(tab.ele('xpath=//*[@id="__next"]/div[8]/div[1]/div/div[2]/div[2]/div/div[1]/div/span/span[2]/span[1]').text.replace('$', ''))
|
|
attr_items.append({
|
|
'url': tab.url,
|
|
'items': [quantity_name],
|
|
'price': price,
|
|
'old_price': old_price,
|
|
'images': color_images
|
|
})
|
|
|
|
if len(attr_items) == 0:
|
|
goods_info.attr = 'S'
|
|
goods_info.attr_items = []
|
|
goods_info.images = main_images
|
|
goods_info.p_lists = []
|
|
else:
|
|
goods_info.attr = 'M'
|
|
goods_info.attr_items = m_data
|
|
for attr_item in attr_items:
|
|
goods_info.p_lists.append(types.GoodsInfo.GoodsInfoP(
|
|
url=attr_item['url'],
|
|
attr_items=attr_item['items'],
|
|
price=attr_item['price'],
|
|
old_price=attr_item['old_price'],
|
|
images=attr_item['images'],
|
|
))
|
|
for image_url in attr_item['images']:
|
|
if image_url not in main_images:
|
|
main_images.append(image_url)
|
|
goods_info.images = main_images
|
|
except Exception as e:
|
|
try:
|
|
tab.close()
|
|
except:
|
|
pass
|
|
raise Exception(e)
|
|
try:
|
|
tab.close()
|
|
except:
|
|
pass
|
|
return goods_info
|