Working On Creating A Test Database. Anyone With Sample Data For Free

I can't imagine any use of test data that won't require a particular type of data, but:
  http://blog.stackoverflow.com/2009/06/stack-overflow-creative-commons-data-dump/
  http://en.wikipedia.org/wiki/Wikipedia_database

You can get formatted, random data with very little effort.
All you have to do is generate random text and numbers for each field, to your target formats and lengths. Then you generate the sql using loops; alternatively, you could add rows to a datatable/dataset while you loop (that way, you can dump xml as well as test-run before db population).
You can be done in an hour.

I am trying to avoid generating text and the likes. I would rather use data that actually complies with contraints and more. Essentially, I need the Database dump to be packaged with the client so instead of generating "JUNK" text, use real text, real phone numbers, addresses, names, etc. Makes it easier to test, and makes the database reusable.

I have been able to extract as much as possible from Adventure works and Northwind. But looking for more. I know essentially looking for data of different kind.  I am talking of data that pertains to schools and students, to real estate data, votes/polling data etc.   I started off wanting to write a sort of text generator, but I would rather have something non dev testers can use in testing their code and also allows for contraint and format checks needed in over 1000 columns in the current database. So I am building a library of sample data that can be used by the DataGen to generate sample data for the test database. So when QA testers try to do testing, they can easily work with it without just seeing JUNK all over the place they can barely make real sense of.