Autoapplication directive does not document source code

[QTBdev]

Hello,

Autoapplication directive does not document the application source code, as only the metadata files are read by the parse_mlappfile function.

I understand that this limitation is most probably due to the fact that the source code is stored in a .mat file inside the .mlapp archive, instead of the usual .m file, and that reading a .mat file outside matlab is not a simple task.

Still I wanted to make this info show up on the project somewhere because I spent a good amount of time trying to understand why it wouldn't autodocument my application.

[joeced]

Hi

I haven't used App Designer myself, so I would like to know what extra information that you need from the file? Could you share a simple App and what you expect to be documented?

[QTBdev]

Hello,

Of course, you will find attached an archive containing :

• A simple application example
• A class created by copy-pasting the whole application source code into a brand new .m file
• And the sphinx generated documentation of both of them

I expected autoapplication to document the application just like autoclass documents the class created with the exact same source code, but the parse_mlappfile function does not look into the file that contains the application source code autoapplicationTest.zip .

[joeced]

Thanks for the example! I think I found a way to get the documentation you expected. As the mlapp file is just a ZIP-file, I examined every file in there.

In <root>/matlab/document.xml there is a verbatim copy of the code that is used in the App. We can extract that, and parse it, and then we should be able to get the documentation you want.

[QTBdev]

That would be great ! I knew the application source code was stored in <root>/appdesigner/appModel.mat, but I didn't realise a copy was also available in this file

[joeced]

I took a look at it, and frankly I'm not sure how to proceed exactly.

In the parse_mlappfile function in mat_types.py I can extract the MATLAB code with:

with ZipFile(mlappfile, "r") as mlapp:
    document = ET.fromstring(mlapp.read("matlab/document.xml"))

documentNs = {"w": "http://schemas.openxmlformats.org/wordprocessingml/2006/main"}
documentDesc = document.find(".//w:t", documentNs)

Then documentDesc will contain a verbatim copy of your MATLAB code in the app. However, most of this is private methods, even the callbacks. So you would end up with a ton of noise, and very little information.

If anyone has a good idea on how to present this, then please provide a PR.