start_sync_amz.py 15 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257258259260261262263264265266267268269270271272273274275276277278279280281282283284285286
  1. import time
  2. import pandas as pd
  3. from retry import retry
  4. from sync_amz_data.settings import LOG_CONF
  5. import logging.config
  6. logging.config.dictConfig(LOG_CONF)
  7. from sync_amz_data.public.amz_ad_client import shop_infos
  8. from sync_amz_data.DataTransform.Data_ETL import Common_ETLMethod,SP_ETL,SB_ETL,SD_ETL
  9. from apscheduler.schedulers.blocking import BlockingScheduler
  10. from datetime import datetime,timedelta
  11. def pre_func(AWS_CREDENTIALS,date_):
  12. conn = SB_ETL(**AWS_CREDENTIALS).clickhouse_connect()
  13. sb_report = SB_ETL(**AWS_CREDENTIALS)
  14. sb_report.reportV3_campaign_sbCampaigns_ETL(conn, params={"startDate": date_, "endDate": date_,
  15. "date": date_.replace("-", "")})
  16. sb_report.reportV3_adGroup_sbAdGroup_ETL(conn, params={"startDate": date_, "endDate": date_,
  17. "date": date_.replace("-", "")})
  18. sb_report.reportV3_adGroup_sbAdGroup_ETL(conn, params={"startDate": date_, "endDate": date_,
  19. "date": date_.replace("-", "")})
  20. sb_report.reportV3_sbCampaignPlacement_ETL(conn, params={"startDate": date_, "endDate": date_,
  21. "date": date_.replace("-", "")})
  22. sb_report.reportV3_sbTargeting_ETL(conn,
  23. params={"startDate": date_, "endDate": date_, "date": date_.replace("-", "")})
  24. sb_report.reportV3_sbSearchTerm_ETL(conn,
  25. params={"startDate": date_, "endDate": date_, "date": date_.replace("-", "")})
  26. sb_report.reportV3_sbAds_ETL(conn, params={"startDate": date_, "endDate": date_, "date": date_.replace("-", "")})
  27. sb_report.reportV3_purchasedAsinRecord_ETL(conn, params={"startDate": date_, "endDate": date_,
  28. "date": date_.replace("-", "")})
  29. sb_report.reportV3_purchasedAsinRecord_ETL(conn, params={"startDate": date_, "endDate": date_,
  30. "date": date_.replace("-", "")})
  31. sb_report.reportV2_campaignsRecord_ETL(conn, params={"startDate": date_, "endDate": date_,
  32. "date": date_.replace("-", "")})
  33. sb_report.reportV2_campaignsVideo_ETL(conn,
  34. params={"startDate": date_, "endDate": date_, "date": date_.replace("-", "")})
  35. sb_report.reportV2_adGroupsRecord_ETL(conn,
  36. params={"startDate": date_, "endDate": date_, "date": date_.replace("-", "")})
  37. sb_report.reportV2_adGroupsVideo_ETL(conn,
  38. params={"startDate": date_, "endDate": date_, "date": date_.replace("-", "")})
  39. sb_report.reportV2_adsRecord_ETL(conn,
  40. params={"startDate": date_, "endDate": date_, "date": date_.replace("-", "")})
  41. sb_report.reportV2_adsVideo_ETL(conn, params={"startDate": date_, "endDate": date_, "date": date_.replace("-", "")})
  42. sb_report.reportV2_keywordsRecord_ETL(conn,
  43. params={"startDate": date_, "endDate": date_, "date": date_.replace("-", "")})
  44. sb_report.reportV2_keywordsVideo_ETL(conn,
  45. params={"startDate": date_, "endDate": date_, "date": date_.replace("-", "")})
  46. sb_report.reportV2_placementRecord_ETL(conn, params={"startDate": date_, "endDate": date_,
  47. "date": date_.replace("-", "")})
  48. sb_report.reportV2_placementVideo_ETL(conn,
  49. params={"startDate": date_, "endDate": date_, "date": date_.replace("-", "")})
  50. sb_report.reportV2_searchtermsRecord_ETL(conn, params={"startDate": date_, "endDate": date_,
  51. "date": date_.replace("-", "")})
  52. sb_report.reportV2_searchtermsVideo_ETL(conn, params={"startDate": date_, "endDate": date_,
  53. "date": date_.replace("-", "")})
  54. sb_report.reportV2_targetsRecord_ETL(conn,
  55. params={"startDate": date_, "endDate": date_, "date": date_.replace("-", "")})
  56. sb_report.reportV2_targetsVideo_ETL(conn,
  57. params={"startDate": date_, "endDate": date_, "date": date_.replace("-", "")})
  58. conn.close()
  59. time.sleep(5)
  60. conn = SP_ETL(**AWS_CREDENTIALS).clickhouse_connect()
  61. sp_report = SP_ETL(**AWS_CREDENTIALS)
  62. sp_report.reportV3_campaign_spCampaignsETL(conn, params={"startDate": date_, "endDate": date_,
  63. "date": date_.replace("-", "")})
  64. sp_report.reportV3_adGroup_spCampaignsETL(conn, params={"startDate": date_, "endDate": date_,
  65. "date": date_.replace("-", "")})
  66. sp_report.reportV3_campaignPlacement_spCampaignsETL(conn, params={"startDate": date_, "endDate": date_,
  67. "date": date_.replace("-", "")})
  68. sp_report.reportV3_targeting_spTargetingETL(conn, params={"startDate": date_, "endDate": date_,
  69. "date": date_.replace("-", "")})
  70. sp_report.reportV3_searchTerm_spSearchTermETL(conn, params={"startDate": date_, "endDate": date_,
  71. "date": date_.replace("-", "")})
  72. sp_report.reportV3_advertiser_spAdvertisedProductETL(conn, params={"startDate": date_, "endDate": date_,
  73. "date": date_.replace("-", "")})
  74. sp_report.reportV3_asin_spPurchasedProductETL(conn, params={"startDate": date_, "endDate": date_,
  75. "date": date_.replace("-", "")})
  76. conn.close()
  77. time.sleep(5)
  78. conn = SD_ETL(**AWS_CREDENTIALS).clickhouse_connect()
  79. sd_report = SD_ETL(**AWS_CREDENTIALS)
  80. sd_report.reportV3_campaign_sdCampaigns_ETL(conn, params={"startDate": date_, "endDate": date_,
  81. "date": date_.replace("-", "")})
  82. sd_report.reportV3_campaignMT_sdCampaigns_ETL(conn, params={"startDate": date_, "endDate": date_,
  83. "date": date_.replace("-", "")})
  84. sd_report.reportV3_adgroup_sdAdGroup_ETL(conn, params={"startDate": date_, "endDate": date_,
  85. "date": date_.replace("-", "")})
  86. sd_report.reportV3_adgroupMT_sdAdGroup_ETL(conn, params={"startDate": date_, "endDate": date_,
  87. "date": date_.replace("-", "")})
  88. sd_report.reportV3_targeting_sdTargeting_ETL(conn, params={"startDate": date_, "endDate": date_,
  89. "date": date_.replace("-", "")})
  90. sd_report.reportV3_targetingMT_sdTargeting_ETL(conn, params={"startDate": date_, "endDate": date_,
  91. "date": date_.replace("-", "")})
  92. sd_report.reportV3_asin_sdPurchasedProduct_ETL(conn, params={"startDate": date_, "endDate": date_,
  93. "date": date_.replace("-", "")})
  94. sd_report.reportV3_advertiser_sdAdvertisedProduct_ETL(conn, params={"startDate": date_, "endDate": date_,
  95. "date": date_.replace("-", "")})
  96. # sd_report.reportV2_campaignsRecord_t2_ETL(conn,params=para)
  97. # sd_report.reportV2_campaignsRecord_t3_ETL(conn,params=para)
  98. # sd_report.reportV2_adGroupsRecord_t2_ETL(conn,params=para)
  99. # sd_report.reportV2_adGroupsRecord_t3_ETL(conn,params=para)
  100. # sd_report.reportV2_asins_t2_ETL(conn,params=para)
  101. # sd_report.reportV2_asins_t3_ETL(conn,params=para)
  102. # sd_report.reportV2_productAds_t2_ETL(conn,params=para)
  103. # sd_report.reportV2_productAds_t3_ETL(conn,params=para)
  104. # sd_report.reportV2_targets_t2_ETL(conn,params=para)
  105. # sd_report.reportV2_targets_t3_ETL(conn,params=para)
  106. # sd_report.reportV2_campaign_matchedTarget_t2_ETL(conn,params=para)
  107. # sd_report.reportV2_campaign_matchedTarget_t3_ETL(conn,params=para)
  108. # sd_report.reportV2_adGroups_matchedTarget_t2_ETL(conn,params=para)
  109. # sd_report.reportV2_adGroups_matchedTarget_t3_ETL(conn,params=para)
  110. # sd_report.reportV2_targets_matchedTarget_t2_ETL(conn,params=para)
  111. # sd_report.reportV2_targets_matchedTarget_t3_ETL(conn,params=para)
  112. conn.close()
  113. time.sleep(5)
  114. def amz_report(AWS_CREDENTIALS,para=None):
  115. count = 1
  116. list_date = [((datetime.today()+timedelta(days=i)).date()).isoformat() for i in range(-2,-8,-1)]
  117. # conn = SB_ETL(**AWS_CREDENTIALS).clickhouse_connect()
  118. # conn.command('')
  119. if para is not None:
  120. list_date = [para['startDate']]
  121. # print(list_date)
  122. delete_list = [
  123. 'SB_sbAdGroup_adGroupV3',
  124. 'SB_sbAds_adsV3',
  125. 'SB_sbCampaigns_campaignV3',
  126. 'SB_sbCampaigns_placementV3',
  127. 'SB_sbPurchasedProduct_asinV3',
  128. 'SB_sbSearchTerm_searchTermV3',
  129. 'SB_sbTargeting_targetingV3',
  130. 'SD_advertiser_sdAdvertisedProductV3',
  131. 'SD_asin_sdPurchasedProductV3',
  132. 'SD_sdAdGroup_adGroupMatchedTargetV3',
  133. 'SD_sdAdGroup_adGroupV3',
  134. 'SD_sdCampaigns_campaignMatchedTargetV3',
  135. 'SD_sdCampaigns_campaignV3',
  136. 'SD_targeting_sdTargetingMatchedTargetV3',
  137. 'SD_targeting_sdTargetingV3',
  138. 'SP_spAdvertisedProduct_advertiserV3',
  139. 'SP_spCampaigns_adGroupV3',
  140. 'SP_spCampaigns_campaignV3',
  141. 'SP_spCampaigns_placementV3',
  142. 'SP_spPurchasedProduct_asinV3',
  143. 'SP_spSearchTerm_searchTermV3',
  144. 'SP_spTargeting_targetingV3']
  145. # conn = SP_ETL(**AWS_CREDENTIALS).clickhouse_connect()
  146. # for i in delete_list:
  147. # conn.command(f"delete from AmazonReport.{i} where date='{(datetime.today()-timedelta(days=8)).date().isoformat()}' and profileId='{AWS_CREDENTIALS['profile_id']}'")
  148. for date_ in list_date:
  149. print(date_)
  150. print(date_.replace("-", ""))
  151. # for i in delete_list:
  152. # conn.command(f"delete from AmazonReport.{i} where date='{list_date[-2]}' and profileId='{AWS_CREDENTIALS['profile_id']}'")
  153. try:
  154. refresh_token = shop_infos(AWS_CREDENTIALS['profile_id'])['refresh_token']
  155. AWS_CREDENTIALS['refresh_token'] = refresh_token
  156. pre_func(AWS_CREDENTIALS,date_)
  157. except Exception as e:
  158. print(e)
  159. pre_func(AWS_CREDENTIALS,date_)
  160. if datetime.today().weekday() in [2,6]:
  161. print('execute func2')
  162. try:
  163. date_list = func_2(delete_list)
  164. func_3(date_list)
  165. except Exception as e:
  166. logging.warning(e)
  167. date_list = func_2(delete_list)
  168. func_3(date_list)
  169. def func_2(delete_list):
  170. conn = SP_ETL(**AWS_CREDENTIALS).clickhouse_connect()
  171. date_list = []
  172. for table_name in delete_list:
  173. check_date = (datetime.today() + timedelta(days=14)).date().isoformat()
  174. try:
  175. locals()[table_name] = conn.query_df(
  176. f"select date,SUM(impressions) as impressions from AmazonReport.`{table_name}` where date >='{check_date}' group by date,profileId")
  177. except:
  178. try:
  179. locals()[table_name] = conn.query_df(
  180. f"select date,SUM(salesOtherSku1d) as salesOtherSku1d from AmazonReport.`{table_name}` date >='{check_date}' group by date,profileId")
  181. except:
  182. try:
  183. locals()[table_name] = conn.query_df(
  184. f"select date,SUM(salesBrandHalo) as salesBrandHalo from AmazonReport.`{table_name}` date >='{check_date}' group by date,profileId")
  185. except:
  186. locals()[table_name] = conn.query_df(
  187. f"select date,SUM(sales14d) as sales14d from AmazonReport.`{table_name}` date >='{check_date}' group by date,profileId")
  188. locals()[table_name].columns = ['date', 'refer_']
  189. locals()[table_name]['refer_'] = locals()[table_name]['refer_'].astype('float64')
  190. todayMinus2 = (datetime.today().date() + timedelta(days=-2))
  191. reback_21days = (todayMinus2 + timedelta(days=-8)).isoformat()
  192. df = pd.date_range(start=reback_21days, end=todayMinus2.isoformat(), freq='1d')
  193. df1 = pd.DataFrame(df, columns=['date_'])
  194. merge_df = df1.merge(locals()[table_name], left_on=['date_'], right_on=['date'], how='left')
  195. rel = merge_df.query("date_!=date or refer_<1 or refer_.isna()")['date_'].map(
  196. lambda x: x.date().isoformat()).tolist()
  197. print(rel, date_list)
  198. if len(rel) > 0:
  199. for date in rel:
  200. date_list.extend([date])
  201. conn = SP_ETL(**AWS_CREDENTIALS).clickhouse_connect()
  202. conn.command(
  203. f"delete from AmazonReport.`{table_name}` where date='{date}' and profileId='{AWS_CREDENTIALS['profile_id']}'")
  204. return date_list
  205. def func_3(date_list):
  206. if len(date_list) > 0:
  207. print(list(set(date_list)))
  208. for date_retry in list(set(date_list)):
  209. try:
  210. pre_func(AWS_CREDENTIALS, date_retry)
  211. except:
  212. pre_func(AWS_CREDENTIALS, date_retry)
  213. if __name__ == '__main__':
  214. # AccountTask("3006125408623189").do({"record": "portfolios"})
  215. AWS_CREDENTIALS = {
  216. 'lwa_client_id': 'amzn1.application-oa2-client.ebd701cd07854fb38c37ee49ec4ba109',
  217. 'refresh_token': "Atzr|IwEBIL4ur8kbcwRyxVu_srprAAoTYzujnBvA6jU-0SMxkRgOhGjYJSUNGKvw24EQwJa1jG5RM76mQD2P22AKSq8qSD94LddoXGdKDO74eQVYl0RhuqOMFqdrEZpp1p4bIR6_N8VeSJDHr7UCuo8FiabkSHrkq7tsNvRP-yI-bnpQv4EayPBh7YwHVX3hYdRbhxaBvgJENgCuiEPb35Q2-Z6w6ujjiKUAK2VSbCFpENlEfcHNsjDeY7RCvFlwlCoHj1IeiNIaFTE9yXFu3aEWlExe3LzHv6PZyunEi88QJSXKSh56Um0e0eEg05rMv-VBM83cAqc5POmZnTP1vUdZO8fQv3NFLZ-xU6e1WQVxVPi5Cyqk4jYhGf1Y9t98N654y0tVvw74qNIsTrB-8bGS0Uhfe24oBEWmzObvBY3zhtT1d42myGUJv4pMTU6yPoS83zhPKm3LbUDEpBA1hvvc_09jHk7vUEAuFB-UAZzlht2C1yklzQ",
  218. 'lwa_client_secret': 'cbf0514186db4df91e04a8905f0a91b605eae4201254ced879d8bb90df4b474d',
  219. 'profile_id': "3006125408623189"
  220. }
  221. try:
  222. refresh_token = shop_infos(AWS_CREDENTIALS['profile_id'])['refresh_token']
  223. AWS_CREDENTIALS['refresh_token'] = refresh_token
  224. except Exception as e:
  225. print(e)
  226. timezone_ = Common_ETLMethod(**AWS_CREDENTIALS).timeZone()
  227. print(timezone_)
  228. # amz_report(AWS_CREDENTIALS)
  229. sched = BlockingScheduler()
  230. sched.add_job(amz_report,'cron',hour=17,minute=0,second=0,timezone=timezone_, args=(AWS_CREDENTIALS, ))#,params={"startDate":"2023-11-04","endDate":"2023-11-04","date":"20231104"}
  231. sched.start()