Fix #747

[damonmcminn]

#747

If req.url is an empty string, url.parse(req.url).path is null and common.urlJoin will raise a TypeError.

[jcrugzz]

@damonmcminn whats the use case of making req.url an empty string out of curiosity? Thanks for the fix!

[jcrugzz]

fix in 1.9.1

[damonmcminn]

@jcrugzz No problem! I don't believe there is a use case for an empty string... It was an accidental discovery when I changed a regex that is used to strip the first path from req.url:
https://github.com/damonmcminn/api-proxy/commit/7dda2413535c5d5c1be0f574dde8ee69cd827fa7

The stripped path is used to identify what server to proxy the request to (so I can serve multiple APIs off a single validated domain i.e. api.damonmcminn.com) e.g.:
api.damonmcminn.com/nutrition/green-turtle => localhost:50000/green-turtle

The original regex only matched when a trailing slash was present, so req.url would be rewritten thus '/path/' => '/' but as I wanted to match regardless of trailing slash, I was erroneously rewriting '/path' => ''.

I am no longer rewriting req.url to an empty string but provided the fix in case someone else inadvertently does (or wants to).