go-chi
Don't strip trailing slash if it's escaped in StripSlashes middlware
This fixes an edge case where a path ending in a trailing encoded slash %2F is incorrectly stripped and rctx.RoutePath is set to the unescaped Path, resulting in routing errors.
The specific edge case in my project is:
- rctx is nil
- rctx.RoutePath is empty
- r.URL.RawPath contains multiple escaped slashes in various places
- rctx.RoutePath is set to the unescaped values contained in r.URL.Path
- The Mux no longer matches the route
/link_resolver/{url}because all%2Fare now/
A similar issue also exists in RedirectSlashes, I can open a second PR for that as well if you'd like.
@VojtechVitek it seems like the author may have abandoned this PR, but it's needed: the current behavior strips trailing %2F strings from the URL inappropriately. I've tested this code in our project and it fixed the issue. If @nuuls isn't around to kick-off CI I'm happy to open a new PR with this content to do so.
Although this might properly be considered a breaking change since in my experience it changes the routing of plenty of real-world URLs that apps might have depended on.
