上传文件至 ''
Embrace affordable luxury with our [Gold Plated Jewelry Wholesale](https://www.kirinjewelrywholesale.com/collections/plating-plating). Immerse yourself in the radiance of meticulously designed pieces that bring a touch of glamour to your look, celebrating elegance without compromise.main LSJ
parent
2500e17e40
commit
ca8a652b38
@ -0,0 +1,56 @@
|
||||
#coding = utf-8
|
||||
|
||||
import requests
|
||||
import parsel
|
||||
import os
|
||||
|
||||
headers = {
|
||||
"user-agent": "Mozilla/5.0 (Windows NT 10.0; WOW64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/89.0.4389.90 Safari/537.36"
|
||||
}
|
||||
|
||||
def get_onepage(url):
|
||||
response = requests.get(url=url,headers=headers)
|
||||
selector = parsel.Selector(response.text)
|
||||
lis = selector.css('.listdiv ul li')
|
||||
for li in lis:
|
||||
fail_title = li.css('.galleryli_title a ::text').get() # 文件名
|
||||
href = li.css('.galleryli_title a ::attr(href)').get() # 详情页短地址
|
||||
# global href_full
|
||||
href_full = "https://www.fnvshen.com/" + href # 进入详情页的完整地址
|
||||
print("正在准备下载" + href_full)
|
||||
# print('正在下载相册:', fail_title)
|
||||
|
||||
# 创建相册文件夹
|
||||
if not os.path.exists('img\\' + fail_title): # 如果该路径下没有该文件夹
|
||||
os.mkdir('img\\' + fail_title)
|
||||
|
||||
try:
|
||||
for page_url_num in range(1,101):
|
||||
page_url = href_full + "{}.html".format(page_url_num)
|
||||
response_page = requests.get(url=page_url,headers=headers)
|
||||
selector_page = parsel.Selector(response_page.text)
|
||||
imgs = selector_page.css(".gallery_wrapper ul#hgallery img") #获取详情页中单页的对象
|
||||
for img in imgs:
|
||||
img_name = img.css('img ::attr(alt)').get() #获得图片储存名称
|
||||
img_href = img.css('img ::attr(src)').get() #图片下载地址
|
||||
# print(img_name,img_href)
|
||||
suffix = img_href.split('/')[-1] #获取图片下载后缀
|
||||
response_page_1 = requests.get(url=img_href,headers=headers).content
|
||||
with open(f'img\\{fail_title}\\{img_name}' + suffix, mode='wb') as f:
|
||||
f.write(response_page_1)
|
||||
print("保存完成:", img_name)
|
||||
except:
|
||||
print("该页已经是最后一页")
|
||||
continue
|
||||
|
||||
|
||||
# https://www.fnvshen.com/gallery/4.html
|
||||
# https://www.fnvshen.com/gallery/3.html
|
||||
|
||||
for page in range(1,101):
|
||||
print("正在下载第{}页".format(page))
|
||||
url = "https://www.fnvshen.com/gallery/{}.html".format(page)
|
||||
get_onepage(url)
|
||||
|
||||
|
||||
|
@ -0,0 +1,67 @@
|
||||
#coding = utf-8
|
||||
|
||||
import requests
|
||||
import parsel
|
||||
import os
|
||||
|
||||
|
||||
headers = {
|
||||
"User-Agent": "Mozilla/5.0 (Windows NT 10.0; WOW64) AppleWebKit/537.36 (KHTML, like Gecko) Chrome/89.0.4389.90 Safari/537.36"
|
||||
}
|
||||
|
||||
def get_oneclass(url_class):
|
||||
response_page_class = requests.get(url=url_class,headers=headers)
|
||||
selector_class = parsel.Selector(response_page_class.text)
|
||||
# url_class_sorts = selector_class.css('.top_nav ul li ::attr(href)').getall()
|
||||
# for url_class_sort in url_class_sorts:
|
||||
dls = selector_class.css('.chanpin_list dl')
|
||||
for dl in dls:
|
||||
dl_url = dl.css('dt a ::attr(href)').get()
|
||||
global dl_name
|
||||
dl_name = dl.css('dt a ::attr(alt)').get() #创建文件夹名
|
||||
if not os.path.exists('img\\' + dl_name): # 如果该路径下没有该文件夹
|
||||
os.mkdir('img\\' + dl_name)
|
||||
get_onepeple(dl_url)
|
||||
|
||||
def get_onepeple(url):
|
||||
response_page = requests.get(url=url,headers=headers)
|
||||
selector = parsel.Selector(response_page.text)
|
||||
imgs = selector.css('.neirong_body p strong img')
|
||||
for img in imgs:
|
||||
img_url = img.css('img ::attr(src)').get()
|
||||
img_name = img_url.split("/")[-1]
|
||||
# suffix = img_url.split(".")[-1]
|
||||
# print(img_url)
|
||||
|
||||
response_page_1 = requests.get(url=img_url,headers=headers).content
|
||||
with open(f"img\\{dl_name}\\{img_name}", mode='wb') as f:
|
||||
f.write(response_page_1)
|
||||
print("保存完成:", img_name)
|
||||
|
||||
|
||||
url_list_1 = []
|
||||
|
||||
url_aepnu = "http://www.xunfangimg.com/aepnu/list_1.html"
|
||||
url_z7qnv_1 = 'http://www.xunfangimg.com/z7qnv/list_1.html'
|
||||
url_z7qnv_2 = 'http://www.xunfangimg.com/z7qnv/list_2.html'
|
||||
url_zp7sg_1 = 'http://www.xunfangimg.com/zp7sg/list_1.html'
|
||||
url_full_1 = "http://www.xunfangimg.com/"
|
||||
|
||||
url_list_2 = [url_aepnu,url_z7qnv_1,url_z7qnv_2,url_zp7sg_1,url_full_1]
|
||||
|
||||
response_full = requests.get(url=url_full_1,headers=headers)
|
||||
selector_full = parsel.Selector(response_full.text)
|
||||
lis_full = selector_full.css('.top_nav ul li ')
|
||||
for li_full in lis_full:
|
||||
url_class_1 = li_full.css('a ::attr(href)').get()
|
||||
# print(url_class_1)
|
||||
url_list_1.append(url_class_1)
|
||||
|
||||
url_list = url_list_1 + url_list_2
|
||||
# print(url_list)
|
||||
for url in url_list:
|
||||
print(url + "开始爬取" )
|
||||
try:
|
||||
get_oneclass(url)
|
||||
except:
|
||||
print("爬取失败")
|
Loading…
Reference in New Issue