|
@@ -151,15 +151,12 @@ class DataProcessor(object):
|
|
|
"""
|
|
|
video_id, hour_dt = item_
|
|
|
label_info = self.label_data
|
|
|
- if not label_info:
|
|
|
- print(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
|
|
|
- set video_title = '{title}', label = '{label}', daily_dt_str = '{dt_daily}'
|
|
|
- where video_id = '{video_id}'
|
|
|
- ;"""
|
|
|
+ set video_title = '{title}', label = '{label}', daily_dt_str = '{dt_daily}' where video_id = '{video_id}';"""
|
|
|
+ print(insert_sql)
|
|
|
self.client_spider.update(insert_sql)
|
|
|
|
|
|
select_sql = "SELECT video_id, hour_dt_str FROM lightgbm_data where label is NULL and hour_dt_str < '20240327';"
|