I'm somewhat new to RESTful API design practice and looking for help. The data model I'm working with details a top-level object (e.g. Book) containing a list of a nested objects (e.g. Chapters).
Each Chapter is persisted in its own right and has a simple_name field that is set to unique=True (ie. only one Chapter with Chapter.simple_name = "ch_01" may exist).
Scenario: A user POSTs a Book ("myBook_v1") containing Chapters, "ch_01", "ch_02", and "ch_03". The user then edits their book and the next day POSTs "myBook_v2" containing Chapters, "ch_01", "ch_02", "ch_03", and "ch_04". Suppose that Chapters "ch_01", "ch_02", "ch_03" are unchanged from the original POST.
Currently, since simple_name is required to be unique, the second POST does not pass uniqueness validation and an error response is returned to the user.
Question: Would the following implementation fit REST design principles? And, most importantly, is this a good (safe) design choice?
Implementation: Upon Book POST, check each Chapter.simple_name for uniqueness. If Chapter.simple_name already exists in database, skip creation. If it does not exist, create new Chapter. Return complete persisted Book with a mix of newly created Chapters and already existing Chapters. The only criteria for deciding to create new or use existing is whether or not the user-specified simple_name already exists for a Chapter in the database.
