How to delete hyphen and underscore from url string in nginx?

Question

I have been trying to delete ANY/ALL hyphens - and underscores _ from an incoming URL request to my nginx server.

So to be clear, when someone enters a URL as follows:

https://www.example.com/my-name_is-tom

...I need for nginx to rewrite the URL as follows:

https://www.example.com/mynameistom

I am working with the following config:

server {
   listen 80;
   return 301 https://$host$request_uri;
}

server {
    listen 443 ssl;
    server_name top.example.com;

    ssl_certificate     /etc/ssl/top.example.com.crt;
    ssl_certificate_key /etc/ssl/top.example.com.key;

    # set the root
    root /srv/top.exemple.com;
    index index.html;


    location ~ ^/([a-zA-Z0-9=\?\_\-]+)$ {
        rewrite ^(/.*)-(.*)$ $1$2 last;
        rewrite ^(/.*)_(.*)$ $1$2 last;
        rewrite ^/(.*)$ / break;
    }

    location / {
        ssi on;
    }
    # BOSH
    location /http-bind {
        proxy_pass      http://localhost:0000/http-bind;
        proxy_set_header X-Forwarded-For $remote_addr;
        proxy_set_header Host $http_host;
    }
}

...however, I see no re-writing taking place.

1. Maybe I crafted that location rewrite thing wrong?

2. Maybe I need to somehow rewrite the X-Forward-For $remote_addr; ???

Any insight / suggestions would be MUCH appreciated -- I just don't know much about nginx and regexp.

Thank you all in advance for any time and attention.

EDIT/PS. It seems that I need some kind of rule that removes non-alphanumerics from $request_uri. So this:

example.com/my-name-is-tom.html

would be visually re-written in the browser URL field to:

example.com/mynameistomhtml

I realize how totally odd this sounds, but... that's what needs to happen.

Any further insight would be tremendously appreciated. TY!

Answer 1

To change the URL displayed in the client's browser's address field, you need an external redirect:

rewrite ^(.*)[-_](.*)$ $1$2 permanent;

If you need to restrict the scope of the rewrite, for example, so that /http-bind/ is not rewritten, you can make the regex more specific:

rewrite ^(/[^/]*)[-_]([^/]*)$ $1$2 permanent;

Explanation: capture and match leading slash followed by zero or more non-slash characters. Match hyphen or underscore. Capture and match zero or more non-slash characters.

Both rewrites will redirect with an HTTP 301 response, repeatedly, until all of the [-_] are removed.

Place the rewrite before the first location block.

If you place the rewrite inside a location block, ensure that the location matches the range of URIs that the rewrite is expected to rewrite. However, the rewrite rule is already quite specific, so the presence of a location block is fairly redundant.

See this document for more.