python简单爬虫定时推送同花顺直播及荐股至邮箱

• 1、初衷：实践
• 2、技术：python requests Template
• 3、思路：根据直播页面获取评价最高的前十博主，定时爬行最新的消息和实战股票
• 4、思路：python 编辑简单邮件html模板
• 5、难点：邮件html模板的设计，还需要邮箱支持

  爬虫文件

  '''
  -- #coding:utf-8
  import requests
  from lxml import etree
  from sendmail import sendmail
  import sys, time
  from string import Template
  reload(sys)
  sys.setdefaultencoding('utf-8')

  1 定义类

  class thszb:
  thsbz_list = []
  zb_contest_list = []
  stock_list = []
  def __init(self, url):
  1. self.url = url
  2. self.get_ths_bz()
  3. self.get_zb_content_list()

  2 获取博主基本信息

  def get_ths_bz(self):
  1. t_html = etree.HTML(requests.get(self.url).text).xpath('/html/body/div[1]/a')
  2. tlist =[]
  3. for t in t_html:
  4. d ={}
  5. d["id"]= str(t.xpath('@data-statid')[0])[str(t.xpath('@data-statid')[0]).rindex('_')+1:]
  6. d["title"]= str(t.xpath('div[1]/strong/text()')[0].decode('utf-8').encode('GB18030')).strip().strip(
  7. '\n').strip(
  8. '\r')
  9. d["url"]= str(t.xpath('@data-clienturl')[0])[str(t.xpath('@data-clienturl')[0]).rindex('=')+1:]
  10. d["isAgree"]=int(t.xpath('div[3]/p/text()')[0])
  11. d["oper"]=self.get_zb_stock_url(d["url"])
  12. tlist.append(d)
  13. self.ths_bz_list = sorted(tlist, key=lambda s: s["isAgree"], reverse=True)
  def get_zb_content_list(self):
  1. print(self.ths_bz_list[:10])
  2. for bz inself.ths_bz_list[:10]:
  3. t_html = etree.HTML(requests.get(bz["url"]).text).xpath('//*[@id="J_Mlist"]/div')
  4. if len(t_html)>0:
  5. t = t_html[0]
  6. zbtime = t.xpath("p/span/text()")[0]
  7. zbtext = str(t.xpath("p/text()")[0].decode('utf-8').encode('GB18030')).strip().strip('\n').strip('\r')
  8. print(" %s %s:%s"%(bz['oper'],zbtime, zbtext))
  9. self.zb_contest_list.append(
  10. {"gpurl": bz['oper'],"title": bz['title'],"zbtime": zbtime,"zbtext": zbtext})
  11. returnself.zb_contest_list

  3 获取博主实战股票页面

  def get_zb_stock_url(self, bz_url):
  1. html = requests.get(bz_url).text
  2. t_html_gp = etree.HTML(html).xpath('//*[@id="gotracelink"]/@data-iosurl')
  3. if len(t_html_gp)>0:
  4. return t_html_gp[0]

  4 获取博主实战股票信息

  def get_zb_stock(self, gp_url):
  1. stock_list =[]
  2. stock_list.append(
  3. {"code": u"股票编码".decode('utf-8').encode('GB18030'),"name": u"股票名称".decode('utf-8').encode('GB18030'),
  4. "date": u"买入日期".decode('utf-8').encode('GB18030'),
  5. "money": u"盈亏金额".decode('utf-8').encode('GB18030'),
  6. "rate": u"盈利率".decode('utf-8').encode('GB18030')})
  7. t_html = etree.HTML(requests.get(gp_url).text).xpath('//*[@id="infoTpl"]/ul')
  8. if len(t_html)>0:
  9. for t in t_html:
  10. try:
  11. code = t.xpath('li[1]/div[2]/text()')[0]
  12. name = t.xpath('li[1]/div[1]/text()')[0].decode('utf-8').encode('GB18030')
  13. date = str(t.xpath('li[2]/text()')[0]).strip()
  14. money = str(t.xpath('li[3]/text()')[0]).strip()
  15. rate = str(t.xpath('li[4]/text()')[0]).strip()
  16. stock_list.append({"code": code,"name": name,"date": date,"money": money,"rate": rate})
  17. except:
  18. pass
  19. return stock_list

  5 发送邮件

  def send_mail(self):
  1. mymail = sendmail([''])
  2. s =""
  3. tt =Template(mymail.title_template)
  4. tt_gp =Template(mymail.table_template)
  5. for zb inself.zb_contest_list:
  6. gp_s =""
  7. pglist =self.get_zb_stock(zb["gpurl"])
  8. for gp in pglist:
  9. try:
  10. iffloat(gp["money"])>0.0:
  11. gp["isBold"]='style="color: #F00,; font-weight: bold;"'
  12. gp_s = gp_s + tt_gp.substitute(gp)
  13. except:
  14. pass
  15. s = s + str(tt.substitute(zb))+ gp_s +" </table>"
  16. if mymail.send_mail(u'同花顺直播 %s '% time.strftime("%Y-%m-%d %H:%M", time.localtime())," %s"%(s)):
  17. print("send_mail ok!^_^")
  18. else:
  19. print("send_mail fail!~_~")
  '''
  if name == 'main':
  ths = ths_zb("http://t.10jqka.com.cn/m/zhibo/index.html")
  ths.send_mail()
  '''

  发送邮件

  import smtplib
  from email.mime.text import MIMEText

  1 邮件类

  class sendmail:

  2 python模板

  title_template = '''
  1. <divid="J_Mlist">
  2. <strongstyle="color:red;">$title $zbtime </strong>
  3. <div>
  4. <p>
  5. $zbtext
  6. </p>
  7. </div>
  8. </div>
  9. <tablewidth="400"border="1">
  10. '''
  11. table_template = '''
  12. <tr ${isBold}>
  13. <td>${code}</td>
  14. <td>${name}</td>
  15. <td>${date}</td>
  16. <td><spanclass="ping">${money}</span></td>
  17. <td>${rate}</td>
  18. </tr>
  '''
  mailtolist = []
  mailhost = "smtp.126.com"
  mail_user = ""
  mail_pass = ""
  mail_postfix = "126.com"
  def __init(self, mailto_list):
  1. self.mailto_list = mailto_list

  2发送html格式邮件

  def send_mail(self, sub, content):
  1. me =sub+"<"+self.mail_user +"@"+self.mail_postfix +">"
  2. msg =MIMEText(content, _subtype="html", _charset="gb2312")
  3. msg["Subject"]=sub
  4. msg["From"]= me
  5. msg["To"]=";".join(self.mailto_list)
  6. try:
  7. s = smtplib.SMTP()
  8. s.connect(self.mail_host)
  9. s.login(self.mail_user,self.mail_pass)
  10. s.sendmail(me,self.mailto_list, msg.as_string())
  11. s.close()
  12. returnTrue
  13. exceptException, e:
  14. print str(e)
  15. returnFalse

  效果展示