feat: 添加重复拉取时数据上限

handle/pull_sample_data.py

@@ -126,7 +126,8 @@ def pull_by_max(size=100000):
     log.info("开始推送数据...")
 
     with odps_sdk.execute_sql(
-            'select * from out_winhc_data_analysis_pull_data where ds = ' + latest_ds + '').open_reader(
+            'select * from out_winhc_data_analysis_pull_data where ds = ' + latest_ds + ' limit ' + str(
+                size)).open_reader(
         tunnel=True) as reader:
         for record in reader:
             c = get_value(
@@ -160,6 +161,7 @@ def pull_by_max(size=100000):
 
 if __name__ == '__main__':
     log.info(f"input args: {sys.argv}")
+
     if len(sys.argv) >= 2:
         pull_by_max(size=int(sys.argv[1]))
     else: