Section 14.2. Code Objects

14.2. Code Objects

Callables are a crucial part of the Python execution environment, yet they are only one element of a larger landscape. The grander picture consists of Python statements, assignments, expressions, and even modules. These other "executable objects" do not have the ability to be invoked like callables. Rather, these objects are the smaller pieces of the puzzle that make up executable blocks of code called code objects.

At the heart of every callable is a code object, which consists of statements, assignments, expressions, and other callables. Looking at a module means viewing one large code object that contains all the code found in the module. Then it can be dissected into statements, assignments, expressions, and callables, which recurse to another layer as they contain their own code objects.

In general, code objects can be executed as part of function or method invocations or using either the exec statement or eval() BIF. A bird's eye view of a Python module also reveals a single code object representing all lines of code that make up that module.

If any Python code is to be executed, that code must first be converted to byte-compiled code (aka bytecode). This is precisely what code objects are.

They do not contain any information about their execution environment, however, and that is why callables exist, to "wrap" a code object and provide that extra information.

Recall, from the previous section, the udf.func_code attribute for a UDFs? Well, guess what? That is a code object. Or how about the udm.im_func function object for UDMs? Since that is also a function object, it also has its own udm.im_func.func_code code object. So you can see that function objects are merely wrappers for code objects, and methods are wrappers for function objects. You can start anywhere and dig. When you get to the bottom, you will have arrived at a code object.