Python new test framework should be used (Refer this for more detail)

nope not resolved at all. What you did is only use more hardcoded tests: they are all fixed. That's what hard coded means.

Random tests have to be generated... randomly... on the fly. Different for each tests.

And here is what I did: as you can see, I commented out MOST of the conditions and my solution still passes the tests. So you should take it as a base, uncommenting/commenting out the different parts to catch test cases that make it fail for each commented out condition (EDIT: some conditions are crossing each others, though. I have a bit of overchecking, in there)

Then, you will be able to see how you generate the random tests.