How to retain global variables across modules?

Question

I have a module: generate_ruml that has a global variable, foo . When I run the script python3 generate_ruml.py , it works.

However, when I call the function of generate_ruml (function name is: gruml() ) from another module: gruml_cli using python3 gruml_cli.py , the global variable is not accessed and is considered as None while using Trace .

How do I make the line global foo work?

Below is the code for generate_ruml :

foo = None

class GRUML:
    def generate_sequential_function_calls(self):
        """generate sequential function calls
        for tracing source code and plotting sequence diagram.
        """
        # generating sequence diagram for a use-case
        _ = GenerateSequenceDiagram(
            self.driver_path, self.driver_name, self.source_code_path[0])
        spec = importlib.util.spec_from_file_location(
            self.driver_name, self.driver_path)
        global foo
        foo = self.foo
        foo = importlib.util.module_from_spec(spec)
        spec.loader.exec_module(foo)
        tracer = Trace(countfuncs=1, countcallers=1, timing=1)
        tracer.run('foo.{}()'.format(self.driver_function))

def gruml(source_code_path, **kwargs):
    """driver function of GRUML.
    """
    gruml = GRUML()
    print('Generating RUML for source code at: {}'.format(source_code_path))
    gruml.get_source_code_path_and_modules(source_code_path)
    gruml.get_driver_path_and_driver_name(
        kwargs.get('use_case', None),
        kwargs.get('driver_name', None),
        kwargs.get('driver_path', None),
        kwargs.get('driver_function', None),
    )
    gruml.generate_dependency_data()
    if gruml.use_case:
        gruml.generate_sequential_function_calls()

Below is the code for gruml_cli :

from generate_ruml import gruml
def call_generate_ruml(source_code_link, **kwargs):
    """call gruml function, assume that if use_case is given, all other related 
    arguments will be present, so checking for them can be skipped in here.

    Arguments:
        source_code_link {[type]} -- [description]
    """
    source_code_dir = download_source_code(source_code_link)
    gruml(
        source_code_dir,
        **kwargs
    )

call_generate_ruml(source_code_link, use_case=use_case, driver_name=driver_name, driver_path=driver_path, driver_function=driver_function)

This is the error I receive when I call the function from gruml_cli :

Traceback (most recent call last):
  File "gruml-cli.py", line 64, in <module>
    driver_name=driver_name, driver_path=driver_path, driver_function=driver_function)
  File "gruml-cli.py", line 36, in call_generate_ruml
    **kwargs
  File "/Users/aviralsrivastava/dev/generate_uml/generate_ruml.py", line 249, in gruml
    gruml.generate_sequential_function_calls()
  File "/Users/aviralsrivastava/dev/generate_uml/generate_ruml.py", line 212, in generate_sequential_function_calls
    tracer.run('foo.{}()'.format(self.driver_function))
  File "/usr/local/Cellar/python/3.7.7/Frameworks/Python.framework/Versions/3.7/lib/python3.7/trace.py", line 441, in run
    self.runctx(cmd, dict, dict)
  File "/usr/local/Cellar/python/3.7.7/Frameworks/Python.framework/Versions/3.7/lib/python3.7/trace.py", line 450, in runctx
    exec(cmd, globals, locals)
  File "<string>", line 1, in <module>
NameError: name 'foo' is not defined

I also tried to insert the script generate_ruml.py in the sys path but in vain.

Answer 1

It looks like you need to import the global variable as well. Like:

from generate_ruml import gruml, foo

Answer 2

You can either import the variable as

from generate_ruml import gruml, foo

Or you can reference the variable as, throughout your code.

generate_ruml.foo