There is no way to convert a Viewer object into a Browser Object nicely. Adding a class method for this may be useful.

When calling an IterableFunction class to get a value, the returned object doesn't represent a pyvba class type. For instance, browser.Item(1) may return a COM object but not a Browser, or Viewer object.

The unit tests run on a CAD application called CATIA which is licensed software that many users don't have. Use a more universal application - one that Travic-CI or GitHub Actions can run on would be preferable.

The JSON exporter has no method of dealing with duplicate items. These typically come in the form of Function or Error due to the processing of these types. It may be worthwhile tossing them into an array from the git-go.

The docstring documentation for the keyword arguments is severely lacking, especially since there are useful tools for them. Add information the user should know to the docstrings.

The following methods could be useful in the future and may be worth implementing.

def search(self, name: str, exact: bool = False):
    """Return a dictionary in format {item: [paths]} matching the name.

    Search for all instances of a method or object containing a name.

    Parameters
    ----------
    exact: bool
        A flag that searches for exact matches.
    name: str
        The name of the attribute to search for.

    Returns
    -------
    dict
        The results of the search in format {item: [paths]}.
    """
    ...

def goto(self, path: list):
    """Retrieve an object at a given location.

    Parameters
    ----------
    path: list
        The steps to take to arrive to an object.

    Examples
    --------
    `goto(['Bodies','Item','HybridShapes','GetItem'])` yields the 'GetItem' function.

    """
    ...

def view_vba(self) -> str:
    """Returns a string that replicates the VBA tree."""
    ...

The only existent documentation that exists is docstrings within the package. Formal documentation would be beneficial for others. A documentation generator may be useful.

The testing occurs client-side only. Create an action or CI to fix this. #6 must be resolved first to make this possible.

When running the exporter on Microsoft Excel, the code hangs when reading "MergeArea", "EntireColumn" and a few others. The result is a memory error on the recursive function. This issue needs to be investigated further and handled.

Most of the test cases lack depth. Many edge cases aren't covered and many tests lack depth.

For most data-mining applications, the package could run from a command line. Add a feature that allows data to be collected using the command line and arguments.