I tend to look at Github for "database" type of repositories. Actually, I make some searches using www.gitlogs.com to spot the right project for my tech requirements!
For example, when searching for countries, I found umpirsky/country-list but that was to heavy and ended up using commerceguys/intl.
Same goes with mime types, I found this wonderful: jshttp/mime-db that I ended up using in my adrienrn/php-mimetyper
The idea is to take some community driven datasets that is built upon serious sources and supported by more people than just you alone !