--url-ignore doesn't consider full file path

[tbm]

I'm not 100% sure this is a bug, but I think so.

I generate /blog/feed/rss.xml but my pages link to /blog/feed and I specify rss.xml as the DirectoryIndex in Apache, so just /blog/feed/ works.

Obviously html-proofer doesn't know about the redirect, so I added --url-ignore /blog/feed/ (not sure if there's a better solution?).

This works for pages which have a ../blog/feed/ link, but for pages under /blog/, they have ../feed/ links - they don't mention /blog/ since they are relative links, but obviously the link is to /blog/feed/.

So I think that html-proofer should always consider the full path of a link when checking against --url-ignore. So e.g. a ../feed/ link in /blog/foo/index.html should be converted to /blog/feed/ before evaluating --url-ignore.

What do you think?

[gjtorikian]

Yeah, that's a tricky one. It sounds like during the link evaluation, it should check first to see if it's a file path (../) before expanding it, and then evaluate the ignore.

I don't have time at the moment to dig too deep into this, but I wonder if you could use --url-swap to temporarily "rewrite" these URLs? That is, something like --url-swap "../feed:../blog/feed". Also, correct me if I am wrong but wouldn't rewriting all the URLs also work? That is, rather than ../feed, to just make it all absolute paths: /blog/feed/rss.xml?

[tbm]

Also, correct me if I am wrong but wouldn't rewriting all the URLs also work? That is, rather than ../feed, to just make it all absolute paths: /blog/feed/rss.xml

Sure, but the web site generator I use uses relative links.

Anyway, my workaround is --url-ignore /feed/ which works fine, but I think html-proofer should expand file paths (as you suggest).