mirror of
https://git.sr.ht/~cadence/NewLeaf
synced 2024-11-21 23:27:29 +00:00
Remove extraneous " align:start position:0%" on auto-generated captions
This commit is contained in:
parent
1d52fca3a0
commit
be8a2dad5f
@ -1,3 +1,4 @@
|
||||
import re
|
||||
import requests
|
||||
from extractors.video import extract_video
|
||||
from tools.converters import escape_html_textcontent, get_subtitle_api_url
|
||||
@ -20,6 +21,9 @@ def extract_captions_from_dict(captions, *, lang=None, label=None):
|
||||
url = next(caption["second__remoteUrl"] for caption in captions["captions"] if caption["languageCode"] == lang or caption["label"] == label)
|
||||
with requests.get(url) as r:
|
||||
r.raise_for_status()
|
||||
# remove extraneous " align:start position:0%" on timestamps lines on auto-generated captions
|
||||
if (lang and "auto-generated" in lang) or (label and "auto-generated" in label):
|
||||
return re.sub(r"^([0-9:.]+ --> [0-9:.]+).*$", r"\1", r.content.decode("utf8"), flags=re.MULTILINE)
|
||||
return r
|
||||
|
||||
# List of captions directly from youtube, but no automatic
|
||||
|
Loading…
Reference in New Issue
Block a user