京东商城大规模爬虫
2017-04-10 本文已影响2822人
蜗牛仔
一开始看京东商城的商品，发现很多信息都在网页源代码上，以为会比淘宝的大规模爬取简单点，结果被京东欺骗了无数次，整整写了差不多六个小时，真是坑爹啊。先贴上github地址：https://github.com/xiaobeibei26/jingdong
先说下这个网站，首先在首页随便输入一个想爬取的商品类别，观察到一般商品数目都是100页的，除非有些比较稀少的商品，如图
Paste_Image.png
介绍一下网站的分析过程，默认情况下在首页输入一件商品时，出来的搜索页面是只有30件商品的，屏幕的右侧下拉框拉到下面会触发一个ajax的请求，把剩下的30个商品渲染出来，一般每页60个商品里面是有三个左右是广告的，也就是有效商品一般是57个。这里看一下这个AJAX请求，这个是爬取难点
Paste_Image.png
看一看这个请求头，我当时第一个感觉以为很多参数是可以去掉，拿到一个很简便的链接就可以了
当时没注意，删了很多参数直接请求，结果调试了很久，获得的商品在插进数据库去重的时候都是只剩网页的一般，细细观察了一下发现链接虽然不同，请求回来的商品却是一样的，然后我再细细看了看这个ajax请求，鼓捣了好久，最终发现这个URL后面的每个数字都是每一件商品的ID，而这个ID隐藏在第一次刚打开网页时候最初的那些商品里面，如图.........
Paste_Image.png
这里结合ajax请求的参数看，
Paste_Image.png
然后我又从新改掉爬虫逻辑，改代码，又花了两个小时，好惨啊.......
然后终于可以一次提取完整的网页商品了，最后提示一下，京东网页第一页的商品里面页数page是显示1和2的，第二页是3和4，这个有点特殊，最后上一张爬虫主程序图
# -*- coding: utf-8 -*-
import scrapy
from jingdong.items import JingdongItem
class JdSpider(scrapy.Spider):
name = "jd"
allowed_domains = ["www.jd.com"]
start_urls = ['http://www.jd.com/']
search_url1 = 'https://search.jd.com/Search?keyword={key}&enc=utf-8&page={page}'
#search_url2='https://search.jd.com/s_new.php?keyword={key}&enc=utf-8&page={page}&scrolling=y&pos=30&show_items={goods_items}'
search_url2= 'https://search.jd.com/s_new.php?keyword={key}&enc=utf-8&page={page}&s=26&scrolling=y&pos=30&tpl=3_L&show_items={goods_items}'
shop_url ='http://mall.jd.com/index-{shop_id}.html'
def start_requests(self):
key = '裤子'
for num in range(1,100):
page1 = str(2*num-1)#构造页数
page2 = str(2*num)
yield scrapy.Request(url=self.search_url1.format(key=key,page=page1),callback=self.parse,dont_filter = True)
yield scrapy.Request(url=self.search_url1.format(key=key,page=page1),callback=self.get_next_half,meta={'page2':page2,'key':key},dont_filter = True)
#这里一定要加dont_filter = True，不然scrapy会自动忽略掉这个重复URL的访问
def get_next_half(self,response):
try:
items = response.xpath('//*[@id="J_goodsList"]/ul/li/@data-pid').extract()
key = response.meta['key']
page2 =response.meta['page2']
goods_items=','.join(items)
yield scrapy.Request(url=self.search_url2.format(key=key, page=page2, goods_items=goods_items),
callback=self.next_parse,dont_filter=True)#这里不加这个的话scrapy会报错dont_filter，官方是说跟allowed_domains冲突，可是第一个请求也是这个域名，实在无法理解
except Exception as e:
print('没有数据')
def parse(self, response):
all_goods = response.xpath('//div[@id="J_goodsList"]/ul/li')
for one_good in all_goods:
item = JingdongItem()
try:
data = one_good.xpath('div/div/a/em')
item['title'] = data.xpath('string(.)').extract()[0]#提取出该标签所有文字内容
item['comment_count'] = one_good.xpath('div/div[@class="p-commit"]/strong/a/text()').extract()[0]#评论数
item['goods_url'] = 'http:'+one_good.xpath('div/div[4]/a/@href').extract()[0]#商品链接
item['shops_id']=one_good.xpath('div/div[@class="p-shop"]/@data-shopid').extract()[0]#店铺ID
item['shop_url'] =self.shop_url.format(shop_id=item['shops_id'])
goods_id=one_good.xpath('div/div[2]/div/ul/li[1]/a/img/@data-sku').extract()[0]
if goods_id:
item['goods_id'] =goods_id
price=one_good.xpath('div/div[3]/strong/i/text()').extract()#价格
if price:#有写商品评论数是0，价格也不再源代码里面，应该是暂时上首页的促销商品，每页有三四件，我们忽略掉
item['price'] =price[0]
yield item
except Exception as e:
pass
def next_parse(self,response):
all_goods=response.xpath('/html/body/li')
for one_good in all_goods:
item = JingdongItem()
try:
data = one_good.xpath('div/div/a/em')
item['title'] = data.xpath('string(.)').extract()[0] # 提取出该标签所有文字内容
item['comment_count'] = one_good.xpath('div/div[@class="p-commit"]/strong/a/text()').extract()[0] # 评论数
item['goods_url'] = 'http:' + one_good.xpath('div/div[4]/a/@href').extract()[0] # 商品链接
item['shops_id'] = one_good.xpath('div/div[@class="p-shop"]/@data-shopid').extract()[0] # 店铺ID
item['shop_url'] = self.shop_url.format(shop_id=item['shops_id'])
goods_id = one_good.xpath('div/div[2]/div/ul/li[1]/a/img/@data-sku').extract()[0]
if goods_id:
item['goods_id'] = goods_id
price = one_good.xpath('div/div[3]/strong/i/text()').extract() # 价格
if price: # 有写商品评论数是0，价格也不再源代码里面，应该是暂时上首页的促销商品，每页有三四件，我们忽略掉
item['price'] = price[0]
yield item
except Exception as e:
pass
pipline代码如图
class JingdongPipeline(object):
# def __init__(self):
# self.client = MongoClient()
# self.database = self.client['jingdong']
# self.db = self.database['jingdong_infomation']
#
# def process_item(self, item, spider):#这里以每个用户url_token为ID，有则更新，没有则插入
# self.db.update({'goods_id':item['goods_id']},dict(item),True)
# return item
#
# def close_spider(self,spider):
# self.client.close()
def __init__(self):
self.conn = pymysql.connect(host='127.0.0.1',port=3306,user ='root',passwd='root',db='jingdong',charset='utf8')
self.cursor = self.conn.cursor()
def process_item(self, item, spider):
try:#有些标题会重复，所以添加异常
title = item['title']
comment_count = item['comment_count'] # 评论数
shop_url = item['shop_url'] # 店铺链接
price = item['price']
goods_url = item['goods_url']
shops_id = item['shops_id']
goods_id =int(item['goods_id'])
#sql = 'insert into jingdong_goods(title,comment_count,shop_url,price,goods_url,shops_id) VALUES (%(title)s,%(comment_count)s,%(shop_url)s,%(price)s,%(goods_url)s,%(shops_id)s,)'
try:
self.cursor.execute("insert into jingdong_goods(title,comment_count,shop_url,price,goods_url,shops_id,goods_id)values(%s,%s,%s,%s,%s,%s,%s)", (title,comment_count,shop_url,price,goods_url,shops_id,goods_id))
self.conn.commit()
except Exception as e:
pass
except Exception as e:
pass
运行结果如图
Paste_Image.png
运行了几分钟，每页一千条，共爬了几万条裤子，京东的裤子真是多
mysql数据库插入操作
Paste_Image.png
