Validate isURL if it contains a TLD *or* a protocol

[tsx]

tl;dr: There should be a way to validate isURL if it has a TLD or a protocol part, or both. For example:

• http://domain.com → valid
• domain.com → valid
• http://domain → valid
• domain → invalid

Use case: I use isURL validation to determine if a user-provided string is an URL or not. Based on that, I render that string either as a hyperlink or plain text. Being user-friendly, I allow URLs to be typed without protocol, assuming implicit http:// in that case. However, sometimes users want to link to their servers on a local network, referencing them by just a hostname, like "http://localhost". To avoid rendering any single word as a link, I'd like a valid URL to have at least one of URL features: either a TLD (like .com) or a protocol (like "http://").

I know that there are require_tld and require_protocol options, but if I set both to false, it will see any single word like "domain" as a valid URL, which is not the case.

[chriso]

Why not just:

if (validator.isURL(url, {require_tld: true}) ||
    validator.isURL(url, {require_protocol: true}) {
    // render as URL
}

[tsx]

Well this is definitely an option but requires parsing same input twice. Doing twice more work is not always a good idea performance-wise when trying to decide how to display a few thousands pieces of text on a page.

[mncrff]

I am having a similar issue. For me it seems that require_tld: true isn't working at all. The following code block will return true:

validator.isURL('www.something', {
    require_tld: true
});

I am using version 9.1.2, and have tested it both locally (node 8.7.0) and on https://npm.runkit.com/validator

[dsacramone]

@mncrff I am having the same issue... Why is "http://www.ddd" valid when I pass in {require_tld: true} ???

validator.isURL('http://www.ddd', {
require_tld: true
});

that should fail, but doesn't. Any fix or reason why it's returning true?

[dsacramone]

bueller bueller?

[chriso]

@dsacramone read #642 (comment)

[profnandaa]

Well this is definitely an option but requires parsing same input twice. Doing twice more work is not always a good idea performance-wise when trying to decide how to display a few thousands pieces of text on a page.

@tsx - remember that that work still has to be done by the end of the day, either by the library or library's end-user. I don't see any perf gains here.