#coding:utf-8
import requests ,re,json,pandas as pd,time
from selenium import webdriver #selenium2.48.0 支持phantomjs
from lxml import etree
driver = webdriver.Chrome(r"C:\Users\V\Desktop\chromedriver.exe")
driver.get(url='https://search.jd.com/Search?keyword=5g%E6%89%8B%E6%9C%BA&suggest=1.def.0.0&wq=5g%E6%89%8B%E6%9C%BA&page=1&s=1&click=0')
for i in range(100):
# x管水平,y管垂直
js = 'window.scrollTo(0,%s)'%(i*100)
driver.execute_script(js)
time.sleep(0.5)
tree=etree.HTML(driver.page_source)
daliebiao=[]
book_list = tree.xpath('//ul[@class="gl-warp clearfix"]/li') # 选一个标签作为树根,
for book in book_list:
xiaoliebiao = []
name_peizhi = book.xpath('.//a[@target="_blank"]/em/text()')
name_peizhi = ",".join(name_peizhi)
price = book.xpath('.//div[@class="p-price"]/strong/i/text()')[0]
dianpu = book.xpath('.//div[@class="p-shop"]/span/a/text()')
if len(dianpu) == 0:
pass
else:
dianpu = dianpu[0]
pingjiaxiangqing_href = book.xpath('.//div[@class="p-commit"]/strong/a/@href')[0]

pingjiaxiangqing_href = str("https:") + pingjiaxiangqing_href
id = book.xpath('.//@data-sku')[0]
print(id)
if float(price) < 1000.00:
pass
else:
xiaoliebiao.append(name_peizhi)
xiaoliebiao.append(price)
xiaoliebiao.append(dianpu)
xiaoliebiao.append(pingjiaxiangqing_href)
xiaoliebiao.append(id)
daliebiao.append(xiaoliebiao)
print(xiaoliebiao)
print(len(daliebiao))