def update_db():
with sql.connect(db_name) as conn:
for query in queries:
for page in range(1, 4):
print(query, page)
search_page = download_search_page(query, page)
video_list = parse_search_page(search_page)
for video in video_list:
video_page = download_video_page(video['link'])
video_json_data = parse_video_page(video_page)
if 'watch-time-text' not in video_json_data:
continue
p = compute_prediction(video_json_data)
video_id = video_json_data.get('og:video:url', '')
watch_title = video_json_data['watch-title'].replace(
"'", "")
data_front = {
"title": watch_title,
"score": float(p),
"video_id": video_id
}
data_front['update_time'] = time.time_ns()
print(video_id, json.dumps(data_front))
c = conn.cursor()
c.execute(
"INSERT INTO videos VALUES ('{title}', '{video_id}', {score}, {update_time})"
.format(**data_front))
conn.commit()
return True
def update_db():
with sql.connect(db_name) as conn:
for keyword in keywords:
for page in range(1, 4):
search_page = get_data.download_search_page(keyword, page)
videos_list = get_data.parse_search_page(search_page)
for video in videos_list:
video_page = get_data.download_video_page(video['link'])
video_json_data = get_data.parse_video_page(video_page)
if 'watch-time-text' not in video_json_data: # if the video parsed does not have a title, discard
continue
p = compute_prediction(video_json_data)
video_id = video_json_data.get('og:video:url', '')
video_title = video_json_data['watch-title'].replace(
"'", "''")
data_front = {
"title": video_title,
"score": float(p),
"video_id": video_id
}
print(video_id, json.dumps(data_front))
cursor = conn.cursor()
cursor.execute(
"INSERT INTO videos VALUES ('{title}', '{video_id}', {score})"
.format(**data_front))
conn.commit()
return True
def predict_api():
yt_video_id = request.args.get("yt_video_id", default='')
video_page = get_data.download_video_page(
"/watch?v={}".format(yt_video_id))
video_json_data = get_data.parse_video_page(video_page)
if 'watch-time-text' not in video_json_data:
return "not found"
p = ml_utils.compute_prediction(video_json_data)
output = {"title": video_json_data['watch-title'], "score": p}
return json.dumps(output)