1
0
mirror of https://git.sr.ht/~cadence/NewLeaf synced 2024-11-14 12:07:30 +00:00
NewLeaf/tools/extractors.py

13 lines
388 B
Python
Raw Normal View History

2020-08-13 14:20:11 +00:00
import re
import json
r_yt_initial_data = re.compile(r"""^\s*window\["ytInitialData"\] = (\{.*\});\n?$""", re.M)
def extract_yt_initial_data(content):
m_yt_initial_data = re.search(r_yt_initial_data, content)
if m_yt_initial_data:
yt_initial_data = json.loads(m_yt_initial_data.group(1))
return yt_initial_data
else:
raise Exception("Could not match ytInitialData in content")