If you are simply looking for degree day stats to fill in a box for something prescriptive like a LEED analysis, then Jason's response is almost certainly correct and you are then just simply looking to characterize the weather/climate for someone otherwise unfamiliar.

If however your specific phrasing "... for a building" infers that you are instead looking to practically apply a degree-day analysis to a specific building/climate/historical period... Awesome! I'll provide a response aimed at that sort of scenario.

Missing from the prompt is the point/purpose of such an analysis, which is important detail... So I'll provide some such context for sake of example. Let's assume you are performing a utility analysis for an existing facility, and after asserting a base:seasonal load split on your gas/electricity consumption history you are looking to evaluate the strength of a potential correlation between your seasonal energy consumption profiles and the corresponding degree days. The degree to which you can assert the "seasonal" energies align with CDD and HDD will lend confidence to your efforts and allow you to draw up a simple model correlating input weather conditions (in say, a TMY year) to estimate future projected baseline seasonal energy consumption, upon which you might be able to reasonably bound the potential savings associated with a boiler replacement.

To get there, we could just pull some CDD65 and/or HDD50 stats from a table, but those balance points might not be a good assumption for your specific situation. Let's describe our example further:
You know from an energy audit that seasonal gas usage is defined by only boiler consumption, which in turn service only a 2-pipe heating/cooling loop. The 2-pipe loop, as operated by the plant manager, is always in one of 3 states: heating, cooling, or inoperative. We can characterize the "inoperative" state as occurring generally in a temperature band around the building's balance point, where no substantial heating/cooling is necessary to maintain comfort. Never heating/cooling simultaneously. The BAS automated switchover is shot, but the plant operator does a good job of tracking occupant comfort and weather trends, and shared with us a historical log of dates the plant has been shifted between heating, cooling, and swing season states. Combining this data with historical AMY, we can ascertain that the boilers are characteristically in operation only when the daily average OADB is 57F or less. Further, the chillers are active only when daily average OADB is 66F or higher.

NOW we've established the most appropriate, building/climate/system-specific base temperatures upon which to base the stated degree day correllation study. One could alternatively avoid much of this field work and understanding in the systems' operations, having collected the historical weather and utility data, by simply interatively "backing into" a balance point which in turn produces a good fit, but you run the risk of missing any of these important details (like the deliberate non-operation of the plant ... (more)