Invite Deleter Ignores Redirecting Links

[ZomoXYZ]

Links such as goo.gl and bit.ly links are ignored when checking for discord invite links.

The only way I can think to fix this would be to send a GET request to every link sent and follow the redirects (unless there are too many redirects) to see if the final url matches your Regex found in src/vortex/AutoMod.java:57.

[jagrosh]

As cool of a feature it would be to follow redirects on all links,

• it would not only use a lot more resources but also take a lot longer to evaluate if moderation is needed
• it would expose the bot's IP to any link along the path, and thus be susceptible to attacks
• some redirect paths can be insanely long to follow

I think a better solution to this problem is just to blacklist any redirect sites' urls once I add a blacklisting feature.

[ZomoXYZ]

What about only following known shortlinks, such as goo.gl and bit.ly, they are the most common and both have APIs.

Google: https://developers.google.com/url-shortener/v1/getting_started#expand Bitly: http://dev.bitly.com/links.html#v3_expand

Only following these links shouldn't take up too many more resources or time. These websites may show a generic location to where it was from, but not your actual IP. To stop redirect loops, just set a limit.

Here's some pseudocode to only follow Google and Bitly links:

//find every google/bitly link
links = content.match(/(goo.gl|bit.ly)\/[a-z0-9]+/i)

//function to expand a google link
function expandGoogle(link)
    body = GET "https://www.googleapis.com/urlshortener/v1/url?key=ACCESS_TOKEN&shortUrl=" + link
    body = parseJSON(body)
    return body.longUrl

//function to expand a bitly link
function expandBitly(link)
    body = GET "https://api-ssl.bitly.com/v3/expand?access_token=ACCESS_TOKEN&shortUrl="+ link
    body = parseJSON(body)
    return body.data.expand.long_url

//list of links the script has processed so it will not repeat the same link
processedLinks = []

//for each found link
for i = 0 links.length > i i++
    maxTime = 5 //most number of redirects it will look through in each link
    times = 0
    
    //loop for each redirect
    function loop(link)
        if !processedLinks.has(link)
            processedLinks.push(link)
            
            if link.match("goo.gl")
                link = expandGoogle(link)
                times++
                if times < maxTime
                    loop(link)
            else if link.match("bit.ly")
                link = expandBitly(link)
                times++
                if times < maxTime
                    loop(link)
            else if link.match("discord.gg")
                i = links.length //stop for loop
                message.delete() //replace this with whatever function you run for each discord.gg link
    
    loop(links[i])

This shouldn't loop too many times (maximum of 5*(number of unique bitly/google links), and can be cut off early).

A blacklist feature would make this easier, but it could delete links that are innocent.

[jagrosh]

I don't like making a feature half-way, so while I appreciate the design, if I add a way to check links for redirects, it will check all links at a much larger depth. The exception to this might be things such as discord.me and other popular listing sites.