dw_ad_day.py 1.7 KB

1234567891011121314151617181920212223242526272829303132333435363738394041424344454647484950515253545556575859606162636465666768
  1. import logging
  2. from model.DataBaseUtils import MysqlUtils,CkUtils
  3. from model.DateUtils import DateUtils
  4. from model.DingTalkUtils import DingTalkUtils
  5. logging.getLogger().setLevel(logging.WARNING)
  6. import pandas as pd
  7. db = MysqlUtils()
  8. ck = CkUtils()
  9. du=DateUtils()
  10. def run(dt):
  11. sql=f"""SELECT a.dt,a.ad_id,b.ad_name,b.type,'',a.cost,view_count,click_count,follow_count,order_count,order_amount,
  12. title,description,e.channel,pitcher,book,platform,stage,ifnull(image_id,'')
  13. from
  14. ad_cost_day a
  15. left join ad_info b on a.ad_id=b.ad_id
  16. left join adcreative_info c on b.adcreative_id=c.adcreative_id
  17. left join channel_by_account_daily e on b.account_id=e.account_id and a.dt=e.dt
  18. left join channel_info_daily f on e.channel=f.channel and e.dt=f.dt
  19. where a.dt='{dt}' """
  20. data = db.quchen_text.get_data_list(sql)
  21. # print(data)
  22. # 图片链接拼接
  23. li = []
  24. for i in data:
  25. # print(i)
  26. li.extend(i[-1].split(','))
  27. # print(li)
  28. sql3 = f"select image_id,preview_url from image_info where image_id in ({str(set(li))[1:-1]})"
  29. image_di = {}
  30. image_data = db.quchen_text.getData(sql3)
  31. for x in image_data:
  32. image_di[x[0]] = x[1]
  33. # print(image_di)
  34. for i in data:
  35. y = ''
  36. for j in i[-1].split(','):
  37. if image_di.get(j):
  38. y = y + ',' + image_di.get(j)
  39. i.append(y[1:])
  40. # print(data)
  41. db.dm.executeMany("replace into dw_ad_day values (%s,%s,%s,%s,%s,%s,%s,%s,%s,%s,%s,%s,%s,%s,%s,%s,%s,%s,%s,%s)",data)
  42. def day():
  43. try:
  44. run(du.getNow())
  45. except:
  46. DingTalkUtils.send("广告数据清洗失败")
  47. if __name__ == '__main__':
  48. day()