Bug report: "Extract Domains" actually extracts hostnames

[gwittel]

Describe the bug

Extract domains is more like extract hostnames. Please rename it or implement the claimed functionality (or implement both as separate items).

To Reproduce
Steps to reproduce the behavior or a link to the recipe / input used to cause the bug:

1. Add 'Extract Domains' to a recipe
2. Add input like http://www.host.com or http://www.host.co.uk (or leave out the schemes)
3. Output has www.host.com or www.host.co.uk. Those would be the hostnames whereas host.com and host.co.uk are the domains.

Expected behavior

The domain for www.host.com is host.com.

Screenshots

n/a

Desktop (if relevant, please complete the following information):

• Mac
• Firefox
• 68

Additional context
Add any other context about the problem here.

[n1474335]

In this case, 'Domains' is being used to refer to 'Domain names'. Domain names are any name assigned via DNS to represent an IP address. In your example above, com is the Top Level Domain (TLD), host is the second level domain (a subdomain of com), and www is the third level domain (a subdomain of host). The whole thing together, www.host.com, is technically a Fully Qualified Domain Name (FQDN).

A hostname is a label that identifies a networked device. FQDNs are a type of hostname, but a hostname doesn't necessarily have to use the DNS structure. On a local network you might just assign single word hostnames to identify your devices because globally unique, fully qualified DNS is not really required at that scale.

So the 'Extract Domains' operation isn't really looking for hostnames, it's looking for fully qualified domain names, or as much of the domain name as it can find. There is a potential argument for renaming it to 'Extract Domain names', however I think it's reasonably clear from the context that 'Domains' means 'Domain names' in this scenario.

Does that make sense?

[gwittel]

Yes, the FQDN clarification absolutely makes sense. My first intuition given the naming was toward registerable domains (one hop above public suffix). Feel free to close this out unless you want to move this toward a feature request for extracting the registerable name portion.

[n1474335]

I'll close it for now. Programmatically extracting only the publicly registrable domain would be quite difficult.

Consider the domain m.en.host.co.uk. Determining which subdomain is the correct cut off is not trivial and I'm not sure I'm aware of many use cases that would make it worthwhile.