|
@@ -152,9 +152,10 @@ class DataProcessor(object):
|
|
|
:param item_:
|
|
|
"""
|
|
|
video_id, hour_dt = item_
|
|
|
+ print(type(video_id))
|
|
|
label_info = self.label_data
|
|
|
title = read_title(client=self.client, video_id=video_id)
|
|
|
- label, dt_daily = generate_label(video_id, hour_dt, label_info)
|
|
|
+ label, dt_daily = generate_label(str(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}';"""
|
|
|
print(insert_sql)
|
|
@@ -163,7 +164,7 @@ class DataProcessor(object):
|
|
|
select_sql = "SELECT video_id, hour_dt_str FROM lightgbm_data where label = 0 and hour_dt_str < '20240327';"
|
|
|
init_data_tuple = self.client_spider.select(select_sql)
|
|
|
init_list = list(init_data_tuple)
|
|
|
- for item in tqdm(init_list):
|
|
|
+ for item in tqdm(init_list[:1p]):
|
|
|
# print(item)
|
|
|
process_info(item)
|
|
|
time.sleep(0.5)
|