You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Hey @lemon234071, thanks for reporting. I'm not sure how to handle this. Right now, the URL has to be somehow separated from other tokes (e.g. by a preceding space). In your string, the URLs could be detected by using the ASCII characters in the string. Maybe this can be useful to add a special handling for Chinese texts? I would not adapt the current URL regex for English (etc.). https://github.com/jfilter/clean-text/blob/master/cleantext/constants.py#L62
output:
郭麒麟打卡,且听他分享防疫小知识https://www.zhihu.com/qution/319823639哈哈http//t.cn/a67ov8bt哈哈哈哈http://t.c
Expected:
郭麒麟打卡,且听他分享防疫小知识XXX哈哈XXX哈哈哈哈XXX
The text was updated successfully, but these errors were encountered: