might be wrong, but in Europe a bachelors is 3 years? That makes now about right.
I chose cosmology to first order because I really enjoyed working with a particular professor (who was a cosmologist) at my undergrad institution. I stuck with it though because (a) cosmologists get to think about a wide variety of fundamental physical systems, and (b) the complexity of analyzing your experiments is much higher (more interesting) since you can't really do controlled experiments that isolate your variables.* I'm drawn to this for several reasons, not the least of which is that a lot of things that would be really nice to have working models for (like the economy, weather and foreign policy) it's impossible for one reason or another to have controlled experiments as well. If I don't get a job in academia as a physicist (a likely result since I'm not at a top 10 university, and don't plan on playing the multiple post-doc game), the analytical tools I learn in cosmology should have applications in those fields and help make me marketable. Note this last thing really isn't a reason on its own, since I could be marketable doing almost any kind of physics, but not necessarily for the kind of jobs I want to do (for instance, I could be marketable as a condensed matter experimentalist, but probably more for my engineering/solid state science skills, not for my analysis skills). But since I like what I'm doing now, and I know I can get a job by doing it, I don't have a reason to switch.
Convoluted, but you asked a complicated question. As far as picking your own field, think about what courses you enjoyed (were they mathematical in nature? Did they introduce cool properties of matter? Were they hands on?). This might let you know whether you drift more towards one subfield of physics or another; whether it's fundamental (speculative) theory/model building, more focused/phenomenological theorizing, or large or small scale experiments. Where you fit in this continuum can sometimes really help you figure out where you want to be; for instance, the main reason I stayed out of particle physics was the collaborations were so large, and I like to know my collaborators and have an important part to play in the project.
*Probably most true for cosmology/astrophysics, but not totally untrue for things like particle physics, biophysics, etc. Though with these things, you can normally come up with a suite of orthogonal experiments that let you isolate your parameters through decomposition of your parameter space; this is much more complicated in cosmology, since the number of useful signals is a small finite number, and their relation to each other is not as well understood.