Description
When a URL string contains a unicode RTLO (right-to-left override) character, this results in two separate links being generated.
PoC (screenshot, because GitHub would otherwise show the rest of the line as right-to-left):
Note that my terminal ignores the RTLO, but it's there. In the browser (Chrome in this case), this link looks like this:
The left and right part are separate links. When the user clicks on the right part, shown as aisa.mp4
, they are redirected to http://4pm.asia/
instead.
I'm not sure what the ideal solution would be. If autolinker would strip all RTLO characters passed to it, then legitimate text (that's not part of an URL) might break. However, an RTLO should not be part of the actual URL. (Autolinker does basically no URL encoding whatsoever. Maybe that should be changed? An urlencoded RTLO would be handled (and ignored) by the browser.)