alg_recsys_recall_4h_region_trend.py 5.3 KB

123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144
  1. # -*- coding: utf-8 -*-
  2. import traceback
  3. import datetime
  4. from odps import ODPS
  5. from threading import Timer
  6. from utils import RedisHelper, get_data_from_odps, send_msg_to_feishu
  7. from config import set_config
  8. from log import Log
  9. from records_process import records_process
  10. from queue import Queue
  11. from tqdm import tqdm
  12. import threading
  13. config_, _ = set_config()
  14. log_ = Log()
  15. region_name2code: dict = config_.REGION_CODE
  16. redis_helper = RedisHelper()
  17. def worker(queue, executor):
  18. while True:
  19. row = queue.get()
  20. if row is None: # 结束信号
  21. queue.task_done()
  22. break
  23. executor(row)
  24. queue.task_done()
  25. def records_process_for_list(records, executor, max_size=50, num_workers=10):
  26. # 创建一个线程安全的队列
  27. queue = Queue(maxsize=max_size) # 可以调整 maxsize 以控制内存使用
  28. # 设置线程池大小
  29. num_workers = num_workers
  30. # 启动工作线程
  31. threads = []
  32. for _ in range(num_workers):
  33. t = threading.Thread(target=worker, args=(queue, executor))
  34. t.start()
  35. threads.append(t)
  36. # 读取数据并放入队列
  37. for row in tqdm(records):
  38. queue.put(row)
  39. # 发送结束信号
  40. for _ in range(num_workers):
  41. queue.put(None)
  42. # 等待所有任务完成
  43. queue.join()
  44. # 等待所有工作线程结束
  45. for t in threads:
  46. t.join()
  47. def check_data(project, table, partition) -> int:
  48. """检查数据是否准备好,输出数据条数"""
  49. odps = ODPS(
  50. access_id=config_.ODPS_CONFIG['ACCESSID'],
  51. secret_access_key=config_.ODPS_CONFIG['ACCESSKEY'],
  52. project=project,
  53. endpoint=config_.ODPS_CONFIG['ENDPOINT'],
  54. connect_timeout=3000,
  55. read_timeout=500000,
  56. pool_maxsize=1000,
  57. pool_connections=1000
  58. )
  59. try:
  60. t = odps.get_table(name=table)
  61. log_.info(f"检查分区是否存在-【 dt={partition} 】")
  62. check_res = t.exist_partition(partition_spec=f'dt={partition}')
  63. if check_res:
  64. sql = f'select * from {project}.{table} where dt = {partition}'
  65. log_.info(sql)
  66. with odps.execute_sql(sql=sql).open_reader() as reader:
  67. data_count = reader.count
  68. else:
  69. log_.info("表{}分区{}不存在".format(table, partition))
  70. data_count = 0
  71. except Exception as e:
  72. log_.error("table:{},partition:{} no data. return data_count=0:{}".format(table, partition, e))
  73. data_count = 0
  74. return data_count
  75. def get_table_data(project, table, partition):
  76. """获取全部分区数据"""
  77. records = get_data_from_odps(date=partition, project=project, table=table)
  78. data = []
  79. for record in records:
  80. tmp = {}
  81. for col_name in ["region", "videoid_array_sum", "videoid_array_avg"]:
  82. tmp[col_name] = record[col_name]
  83. data.append(tmp)
  84. return data
  85. def process_and_store(row):
  86. region = row["region"]
  87. if region not in region_name2code:
  88. return
  89. videoid_array_sum = row["videoid_array_sum"]
  90. videoid_array_avg = row["videoid_array_avg"]
  91. region_code = region_name2code[region]
  92. key1 = "alg_recsys_recall_4h_region_trend_sum_" + region_code
  93. key2 = "alg_recsys_recall_4h_region_trend_avg_" + region_code
  94. expire_time = 24 * 3600
  95. redis_helper.set_data_to_redis(key1, videoid_array_sum, expire_time)
  96. redis_helper.set_data_to_redis(key2, videoid_array_avg, expire_time)
  97. log_.info("trend-sum写入数据key={},value={}".format(key1, videoid_array_sum))
  98. log_.info("trend-avg写入数据key={},value={}".format(key2, videoid_array_avg))
  99. """
  100. 数据表链接:https://dmc-cn-hangzhou.data.aliyun.com/dm/odps-table/odps.loghubods.alg_recsys_recall_strategy_trend/
  101. """
  102. def h_timer_check():
  103. try:
  104. log_.info(f"开始执行: {datetime.datetime.strftime(datetime.datetime.today(), '%Y%m%d%H')}")
  105. #1 判断数据表是否生产完成
  106. project = "loghubods"
  107. table = "alg_recsys_recall_strategy_trend"
  108. partition = "2023122019"
  109. table_data_cnt = check_data(project, table, partition)
  110. if table_data_cnt == 0:
  111. log_.info("上游数据{}未就绪{},等待...".format(table, partition))
  112. Timer(60, h_timer_check).start()
  113. else:
  114. log_.info("上游数据就绪,count={},开始读取数据表".format(table_data_cnt))
  115. #2 读取数据表
  116. data = get_table_data(project, table, partition)
  117. log_.info("数据处理完成,数据数量={},开始处理和写入redis。".format(len(data)))
  118. #3 写入redis
  119. records_process(data, process_and_store, max_size=10, num_workers=5)
  120. except Exception as e:
  121. log_.error(f"4小时地域-趋势统计数据更新失败, exception: {e}, traceback: {traceback.format_exc()}")
  122. send_msg_to_feishu(
  123. webhook=config_.FEISHU_ROBOT['server_robot'].get('webhook'),
  124. key_word=config_.FEISHU_ROBOT['server_robot'].get('key_word'),
  125. msg_text=f"rov-offline{config_.ENV_TEXT} - 4小时地域-趋势统计数据更新失败\n"
  126. f"exception: {e}\n"
  127. f"traceback: {traceback.format_exc()}"
  128. )
  129. if __name__ == '__main__':
  130. log_.info("文件alg_recsys_recall_4h_region_trend.py:「4小时地域-趋势统计」 开始执行")
  131. h_timer_check()