dw_image_cost_day.py 4.2 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119
  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,b.type,sum(a.cost),sum(view_count),sum(click_count),sum(follow_count),sum(order_count),sum(order_amount),
  12. title,description,book,platform,stage,e.channel,pitcher,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}' and c.is_video=0
  20. group by a.dt,b.type,title,description,book,platform,stage,image_id,e.channel,pitcher
  21. """
  22. # print(sql)
  23. data = db.quchen_text.get_data_list(sql)
  24. # print(data)
  25. # 图片链接拼接
  26. li = []
  27. for i in data:
  28. # print(i)
  29. li.extend(i[-1].split(','))
  30. #TODO:之后如果一天产生的图片过多,可能超过sql的字符限制,
  31. # 之后数据使用hive,来进行数据存储
  32. sql3 = f"select image_id,preview_url,signature,width,height from image_info where image_id in ({str(set(li))[1:-1]})"
  33. image_di = {}
  34. image_data = db.quchen_text.getData(sql3)
  35. for x in image_data:
  36. image_di[x[0]] = (x[1],x[2],x[3],x[4])
  37. # print(image_di)
  38. for i in data:
  39. preview_url = ''
  40. signature = ''
  41. width = ''
  42. height = ''
  43. image_id = ''
  44. for j in i[-1].split(','):
  45. if image_di.get(j):
  46. image_id = image_id + ',' + j
  47. preview_url = preview_url + ',' + image_di.get(j)[0]
  48. signature = signature + ',' + image_di.get(j)[1]
  49. width = width + ',' + str(image_di.get(j)[2])
  50. height = height + ',' + str(image_di.get(j)[3])
  51. else:
  52. image_id = image_id + ',' +j
  53. preview_url = preview_url + ',' + ' '
  54. signature = signature + ',' + ' '
  55. width = width + ',' + '0'
  56. height = height + ',' + '0'
  57. i[-1]=image_id[1:]
  58. i.append(preview_url[1:])
  59. i.append(signature[1:])
  60. i.append(0)
  61. i.append(width[1:])
  62. i.append(height[1:])
  63. # exit(0)
  64. sql_video = f"""SELECT a.dt,b.type,sum(a.cost),sum(view_count),sum(click_count),sum(follow_count),sum(order_count),sum(order_amount),
  65. title,description,book,platform,stage,e.channel,pitcher,ifnull(image_id,''),g.preview_url,g.signature,1,
  66. g.width,g.height
  67. from
  68. ad_cost_day a
  69. left join ad_info b on a.ad_id=b.ad_id
  70. left join adcreative_info c on b.adcreative_id=c.adcreative_id
  71. left join channel_by_account_daily e on b.account_id=e.account_id and a.dt=e.dt
  72. left join channel_info_daily f on e.channel=f.channel and e.dt=f.dt
  73. left join video_info g on c.image_id=g.video_id
  74. where a.dt='{dt}' and c.is_video=1
  75. group by a.dt,b.type,title,description,book,platform,stage,image_id,e.channel,pitcher """
  76. data_video = db.quchen_text.get_data_list(sql_video)
  77. data.extend(data_video)
  78. #进行数据存储
  79. db.dm.execute(f'delete from dw_image_cost_day where dt="{dt}"')
  80. db.dm.executeMany("replace into dw_image_cost_day values (%s,%s,%s,%s,%s,%s,%s,%s,%s,%s,%s,%s,%s,%s,%s,%s,%s,%s,%s,%s,%s)",data)
  81. def hourly():
  82. try:
  83. logging.info('广告数据清洗,开始')
  84. run(du.getNow())
  85. logging.info('广告数据清洗,结束')
  86. except:
  87. DingTalkUtils().send("广告数据清洗失败")
  88. def day():
  89. logging.info('广告数据清洗,开始')
  90. for i in du.getDateLists(du.get_n_days(-10), du.get_n_days(-1)):
  91. # print(i)
  92. run(i)
  93. logging.info('广告数据清洗,结束')
  94. if __name__ == '__main__':
  95. # run('2021-05-18')
  96. for i in du.getDateLists('2021-05-01','2021-05-20'):
  97. print(i)
  98. run(i)