I hope you have heard about Object Oriented Programming (OOP). It's not matter of OOP language in usage! The language makes the things easier, but it's only a predispose - it doesn't make your code object oriented automatically. Actually, you can use procedure oriented programing language (for instance) and follow object orientation (and opposite) - nothing prevents it. Why am I saying this?
You can (if you want) encapsulate functionality about credentials in a type named... let's say "Credentials". This type may contain everything needed like application key, application secret, refresh token, current access token, and expiration moment of the current access token. Also the same type may has one public property letting access to valid access token that you need each time a regular API call is going to be maid. Every time that property gets accessed, the property getter may check if the current access token is still valid (with some reserve - 3~5 mins). If it is, the current access token is passed back immediately - it is just valid. If it is not (the token expiration moment has passed and the token is expired), refresh step is performed (as discussed already) in the getter. Once the refresh call succeeds, the new expiration moment is calculated (current moment plus validity period) and saved accordingly for future use (comparison) together with the just received access token. At the end, getter returns with the new valid (again) access token.
Following this rule will make you code clear and simple.
This is OOP - functionality encapsulation and reuse. 😉
