# -*- coding:utf-8 -*-
import urllib
import json
from pyquery import PyQuery
import sys
reload(sys)
sys.setdefaultencoding('utf-8')
def get_jd(keyword):
doc = PyQuery('https://search.jd.com/Search?keyword='+urllib.quote(keyword))
elements = doc(".gl-i-wrap")
for obj in elements:
product = {}
product["price"] = PyQuery(obj)(".p-price").text()
product["name"] = PyQuery(obj)(".p-name").text()
product["href"] = PyQuery(obj)(".p-name")("a").attr("href")
print str(product).decode('string_escape')
def get_baidu(keyword):
doc = PyQuery('http://www.baidu.com/s?wd=' + urllib.quote(keyword))
elements = doc(".result")
for eme in elements:
obj = json.loads(PyQuery(eme)("div")(".c-tools").attr("data-tools"))
obj["text"] =PyQuery(eme)(".c-abstract").text()
print str(obj).decode('unicode-escape').encode('utf-8')
keyword = 'iphone'
get_jd(keyword)
get_baidu(keyword)
PYTHON 抓去京东,百度的数据
最新推荐文章于 2024-04-20 00:51:59 发布