Redirect /file/ to /file (strip trailing slash)

[duff (Allan Odgaard)]

If I have a page at example.com/about.html I would like to also make it accessible as example.com/about/ but with a redirect (because in the past, it may have used a slash, so there could still be people linking to the version with the trailing slash).

What I have so far is:

@trailing_slash {
    path_regexp dir (.+)/$
    not file {path}/index.html
}
redir @trailing_slash {http.regexp.dir.1}?{query}

Here we check if we have a trailing slash, and furthermore, we check that there is not an index.html file in the directory: I do not like this check, because this check must mirror any other configuration about index pages, e.g. there could be an index.php, index.rss, or similar.

What I really want is to check that {path} is a non-existing directory. But it seems like the file matcher is specifically checking that {path} exists and is a file.

Any suggestions?

Secondly, any non-existing path with a trailing slash will be matched and redirected, so example.com/non-existing/ will produce a redirect to example.com/non-existing (which will then return 404).

What I would like to do is include a check for the file we redirect to, for example something like this:

@trailing_slash {
    path_regexp dir (.+)/$
    file {http.regexp.dir.1}.html # <-- Does not work!
}
redir @trailing_slash {http.regexp.dir.1}?{query}

This uses the capture from the path_regexp matcher to strip the trailing path component from {path}, but it appears that these captures can only be accessed from the directive that uses the matcher.

Is there any way to do a file test on the “parent” of {path}? Or another way to achieve the redirect?

[jfirestorm44 (J)]

Hello @duff. Using the redir with a handle or handle_path should achieve what you want.

Also take a look at

uri (Caddyfile directive) - Caddy Documentation

If I understand you right you would need something like this

domain.com {
        root * /path/to/site
        file_server {
              index index.html                
         }
         redir /about/ /about.html
         handle_path /about/ {
                 root * /path/to/site
                 file_server {
                   index about.html
                 }
        }
}

That should allow you to access about.html via domain.com/about.html and also domain.com/about/.

Not sure if it’s the most efficient way but it works. Caddy should spit a 404 back at you if you try to navigate to a non-existent site.

The handle_path directive is for Caddy v2.1 only. If you haven’t upgraded then you would need to use handle and use the uri directive linked above to strip the prefix.

handle_path (Caddyfile directive) - Caddy Documentation

I hope that helps.

[duff (Allan Odgaard)]

@jfirestorm44

If I understand you right you would need something like this

Thanks, but no, I wanted a general rule that redirects any non-existing path with a trailing slash to the path without the slash, assuming that there exist a html file at that location.

If you look at what I posted, I already managed to create the redirect for all non-existing paths, but it would redirect even if there wasn’t an html file where it redirected to, which is what I wanted to improve (to immediately return 404, instead of 308 followed by 404).

[jfirestorm44 (J)]

ok gotcha. Misunderstood the question.

[francislavoie]

Unfortunately, the order in which matchers are executed is not guaranteed, so you can’t use the result of a regexp in another matcher in the same block. What you can do instead though is chain them using handle. Something like this:

@pathWithSlash path_regexp dir (.+)/$
handle @pathWithSlash {
	@htmlFileExists file {re.dir.1}.html
	redir @htmlFileExists {re.dir.1}?{query}
}

As for your first question, I don’t think there’s any way to check for the existence of directories yet. What you could do is check for all the possible index files that could be served, like:

not file {path}/index.html {path}/index.php {path}/index.rss