1
0
Fork 0
mirror of https://git.sr.ht/~cadence/NewLeaf synced 2025-12-15 08:26:30 +00:00
NewLeaf/tools/extractors.py
2020-12-03 17:00:06 +13:00

12 lines
432 B
Python

import re
import json
r_yt_initial_data = re.compile(r"""(?:^\s*window\["ytInitialData"\]|var ytInitialData) = (\{.+?\});(?:\s*$|</script>)""", re.S + 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")