How is the __name__ variable in a Python module defined?
I'm aware of the standard example: if you execute a module directly then it's __name__ global variable is defined as "__main__". However, nowhere in the documentation can I find a precise description of how __name__ is defined in the general case. The module documentation says...
Within a module, the module's name (as a string) is available as the value of the global variable
__name__.
...but what does it mean by "the module's name"? Is it just the name of the module (the filename with .py removed), or does it include the fully-qualified package name as well?
How is the value of the __name__ variable in a Python module determined? For bonus points, indicate precisely where in the Python source code this operation is performed.
Answer
It is set to the absolute name of the module as imported. If you imported it as foo.bar, then __name__ is set to 'foo.bar'.
The name is determined in the import.c module, but because that module handles various different types of imports (including zip imports, bytecode-only imports and extension modules) there are several code paths to trace through.
Normally, import statements are translated to a call to __import__, which is by default implemented as a call to PyImport_ImportModuleLevelObject. See the __import__() documentation to get a feel for what the arguments mean. Within PyImport_ImportModuleLevelObject relative names are resolved, so you can chase down the name variables there if you want to.
The rest of the module handles the actual imports, with PyImport_AddModuleObject creating the actual namespace object and setting the name key, but you can trace that name value back to PyImport_ImportModuleLevelObject. By creating a module object, it's __name__ value is set in the moduleobject.c object constructor.