fix-url-regex
Alain St-Denis 2023-10-29 10:03:09 -04:00
rodzic add80e0f6c
commit aa351e27e2
1 zmienionych plików z 2 dodań i 1 usunięć

Wyświetl plik

@ -11,7 +11,8 @@ TAG_PATTERN = re.compile(r'(#[\w\-]+)([)\]_!?*%/.,;\s]+\s*|\Z)', re.UNICODE)
# This will match non matching braces. I don't think it's an issue.
MENTION_PATTERN = re.compile(r'(@\{?(?:[\w\-. \u263a-\U0001f645]*; *)?[\w]+@[\w\-.]+\.[\w]+}?)', re.UNICODE)
# based on https://stackoverflow.com/a/6041965
URL_PATTERN = re.compile(r'((?:(?:https?|ftp)://|^|(?<=[("<\s]))+(?:[\w\-]+(?:(?:\.[\w\-]+)+))(?:[\w.,@?^=%&:\/~+#-]*[\w@?^=%&\/~+#-]))',
URL_PATTERN = re.compile(r'((?:(?:https?|ftp)://|^|(?<=[("<\s]))+(?:[\w\-]+(?:(?:\.[\w\-]+)+))'
r'[\w.,;:@?!$()*^=%&/~+\-#]*(?<![:;,).>"]))',
re.UNICODE)
def decode_if_bytes(text):