|
@@ -152,7 +152,7 @@ class DataProcessor(object):
|
|
|
label_info = self.label_data
|
|
|
if not label_info:
|
|
|
print(label_info)
|
|
|
- print(len(label_info))
|
|
|
+ # print(len(label_info))
|
|
|
title = read_title(client=self.client, video_id=video_id)
|
|
|
label, dt_daily = generate_label(video_id, hour_dt, label_info)
|
|
|
insert_sql = f"""UPDATE lightgbm_data
|
|
@@ -161,10 +161,10 @@ class DataProcessor(object):
|
|
|
;"""
|
|
|
self.client_spider.update(insert_sql)
|
|
|
|
|
|
- select_sql = "SELECT video_id, hour_dt_str FROM lightgbm_data where label = 0 and hour_dt_str < '20240327';"
|
|
|
+ select_sql = "SELECT video_id, hour_dt_str FROM lightgbm_data where label is NULL and hour_dt_str < '20240327';"
|
|
|
init_data_tuple = self.client_spider.select(select_sql)
|
|
|
init_list = list(init_data_tuple)
|
|
|
- for item in init_list:
|
|
|
+ for item in tqdm(init_list):
|
|
|
# print(item)
|
|
|
process_info(item)
|
|
|
# with ThreadPoolExecutor(max_workers=10) as Pool:
|