cille玻璃杯，玻璃杯子视频
chanong
|#!usr/bin/env python# -*-coding:utf-8 -*-__author__='Charles'__date__='2018.03.16'import requestsimport jsonimport reimport timeclass Spider():#初始化，记录采集时间def __init__(self ) :self.time=time.strftime(\'%Y-%m-%d %H:%M:%S\', time.localtime())print u'\n',u'开始数据采集',u '\ n本地时间：',self.time#获取数据def getData(self,url):html=requests.get(url).textrequests.adapters.DEFAULT_RETRIES=5result=json.loads(html)data=result['data' ]返回数据#获取最新评论iddef getNew(self):data=self.getData(url='http://www.scuinfo.com/api/posts pageSize=15')New=data[0]['id']return New #提取数据有效数据写入dict，多个项目为listdef getDetail(self):New=self.getNew()container=[]i=1for id in range(131599,New+1):content={}self .url='http://www.scuinfo. com/api/post id='+str(id)data=self.getData(url=self.url)if isinstance(data,list):body=data.values()[7] likeCount=data.values()[ 6]comment=data.values()[0]#关键词为“玻璃”、“杯子”、“秋天”、“观光”pattern=re.compile(u'\u73bb\ u7483|\u676f|\u6454|\ u89c2\u5149',re.S)items=re.search(pattern,body)if items:content['body']=bodycontent['like']=likeCountcontent['comment']=commentprint u'\n', i, u'\n', u'发言：', body, u'\n', u'点赞：', likeCount, u'', u'评论：', commenttime. sleep (0.01)i +=1container.append (content)else:print 'None'print u'\n\n', u' to ', self.time, u', info has total comments about the glass event', i - 1, u'returncontainer#获取总数评论数#按照点赞数从大到小对评论进行排序，得到热门评论的前100条def getSort(self):container=self.getDetail()print u '\n ' ,u'输出最受欢迎的前100条评论项目如下： 'container.sort(key=lambda k:k.get('comment',0))container.sort(key=lambda k:k.get('like' ,0), reverse=True)forindex,r in enumerate (容器):print u'\n\n序列号：',index+1, u'\n说出：',r['body'] ,u'\nLike: ' ,r['like'],u'comment' ,r['comment']spider=Spider()spider.getSort()