Improve URL matching

[Johannes-Schneider]

Currently, URL recognition is implemented using an error-prone regex.
This should be improved in order to increase reliability and precision.

A good starting point for this issue would be: https://github.com/john-kurkowski/tldextract