142 lines
4.9 KiB
Python
142 lines
4.9 KiB
Python
from core import spiders, types
|
||
from utils import formats, messages
|
||
|
||
class SpiderModule(spiders.Spiders):
|
||
project_name = "friskybusiness"
|
||
|
||
def __init__(self):
|
||
super().__init__(
|
||
self.project_name,
|
||
bitch = 30,
|
||
worker_num = 5,
|
||
switch_clash = False,
|
||
reflush_browser = False
|
||
)
|
||
self.tab.get('https://friskybusiness.sg/')
|
||
# input('wait:')
|
||
|
||
def get_category_urls(self) -> list[str]:
|
||
category_urls = []
|
||
|
||
one_menu_eles = self.tab.eles('xpath=/html/body/header/height-observer/x-header/nav[2]/ul/li')
|
||
for one_index, one_menu_ele in enumerate(one_menu_eles[:-1]):
|
||
|
||
one_link_ele = one_menu_ele.ele('@tag()=summary')
|
||
one_url = one_link_ele.attr('data-follow-link')
|
||
one_name = one_link_ele.text.replace('/', '-')
|
||
category_urls.append(f"{one_url}#{one_name}")
|
||
messages.sendInfo(one_name)
|
||
|
||
if one_index == 0 or one_index == 6:
|
||
two_menu_eles = one_menu_ele.eles('xpath=mega-menu-disclosure/details/div/div/a')
|
||
else:
|
||
two_menu_eles = one_menu_ele.eles('xpath=mega-menu-disclosure/details/div/ul/li', timeout=1)
|
||
for two_index, two_menu_ele in enumerate(two_menu_eles):
|
||
if one_index == 0 or one_index == 6:
|
||
two_link_ele = two_menu_ele
|
||
else:
|
||
two_link_ele = two_menu_ele.ele('@tag()=a')
|
||
two_url = two_link_ele.attr('href')
|
||
two_name = two_link_ele.text.replace('/', '-').replace(',', ',').replace('#', ' ')
|
||
category_urls.append(f"{two_url}#{one_name}/{two_name}")
|
||
|
||
try:
|
||
three_menu_eles = two_menu_ele.eles(f'xpath=div/ul/li')
|
||
except:
|
||
continue
|
||
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('/', '-').replace(',', ',').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]:
|
||
page = 1
|
||
self.tab.get(f"{category_url}?page={page}")
|
||
goods_urls = []
|
||
|
||
#####获取页数####
|
||
|
||
#################
|
||
|
||
goods_urls = []
|
||
while True:
|
||
self.tab.scroll.to_bottom()
|
||
try:
|
||
goods_eles = self.tab.eles('#:snize-product-', timeout=5)
|
||
for goods_ele in goods_eles:
|
||
goods_url = f"{goods_ele.ele('@tag()=a').attr('href')}"
|
||
goods_urls.append(goods_url)
|
||
if len(goods_eles) == 0:
|
||
break
|
||
except:
|
||
break
|
||
page += 1
|
||
self.tab.get(f"{category_url}?page={page}")
|
||
|
||
messages.sendInfo(f"{len(goods_urls)} 个")
|
||
|
||
return goods_urls
|
||
|
||
def get_goods_info(self, url) -> types.GoodsInfo:
|
||
def get_image_urls():
|
||
image_urls = []
|
||
image_eles = tab.eles('xpath=/html/body/main/section[1]/div/div/product-rerender/div/product-gallery/safe-sticky/product-gallery-navigation/button')
|
||
|
||
for image_ele in image_eles[-1:]:
|
||
image_url = image_ele.ele('xpath=img').attr('src')
|
||
image_urls.append(image_url)
|
||
return image_urls
|
||
|
||
tab = self.browser.new_tab(url)
|
||
tab.set.window.max()
|
||
spu = formats.url_to_spu(url)
|
||
|
||
try:
|
||
h1_ele = tab.ele('@tag()=h1')
|
||
brand = ''
|
||
title = h1_ele.text
|
||
|
||
desc = ""
|
||
try:
|
||
desc = formats.clean_html(tab.ele('.liquid').html)
|
||
except:
|
||
pass
|
||
|
||
price_list = tab.ele('.price-list price-list--product')
|
||
price = price_list.ele('@tag()=sale-price').ele('.money').text.replace('$', '').replace(',', '')
|
||
try:
|
||
old_price = price_list.ele('@tag()=compare-at-price').ele('.money').text.replace('$', '').replace(',', '')
|
||
except:
|
||
old_price = price
|
||
|
||
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,
|
||
)
|
||
|
||
goods_info.attr = 'S'
|
||
goods_info.attr_items = []
|
||
goods_info.images = main_images
|
||
goods_info.p_lists = []
|
||
except Exception as e:
|
||
try:
|
||
tab.close()
|
||
except:
|
||
pass
|
||
raise Exception(e)
|
||
try:
|
||
tab.close()
|
||
except:
|
||
pass
|
||
return goods_info
|