Handle PPTX shapes where position is None

[richardye101]

My previous PR for sorting powerpoint shapes relied on the top and left attributes of the pptx shapes, however I didn't account for shapes having None as their top, left, height or width attributes. We'll need to think about what to do with these shapes, as sometimes they can still contain text, however they don't exist on the slide.

Currently, I've filtered out shapes with None in their positional attributes, however I'm open to changing how they're dealt with.

I think the options are:

1. Parse them into markdown in a "Hidden Text" section under the Notes
2. Ignore these shapes as I have written, as they don't appear on the powerpoint anyway so anyone reading it wouldn't know that text exists

[afourney]

Thanks for looking into this!

[afourney]

It looks like there is a stack overflow/infinite recursion error now on the test file.

Moreover, I think that skipping shapes is problematic -- is it possible those shapes are tables, pictures, etc?

Maybe have the sort key be the tuple: (float('inf') if shape.top is None else shape.top, float('inf') if shape.left is None else shape.left),

so that shapes with missing coordinates always get listed last (but because it's a stable sort, they appear in the order they were indexed in the file)

[richardye101]

From my research, the only shapes with None in their top,left,height,width attributes are of type pptx.shapes.placeholder.SlidePlaceholder. I'm not sure why the sort key would result in a stack overflow error though...

To your point about listing the shapes last: It seems almost all the shapes with missing coordinates are slide titles using placeholder shapes inserted using a slide template. I've seen some instances where a placeholder for slide content is also missing coordiantes. Based on that, i would actually be inclined to list them first

[afourney]

From my research, the only shapes with None in their top,left,height,width attributes are of type pptx.shapes.placeholder.SlidePlaceholder. I'm not sure why the sort key would result in a stack overflow error though...

To your point about listing the shapes last: It seems almost all the shapes with missing coordinates are slide titles using placeholder shapes inserted using a slide template. I've seen some instances where a placeholder for slide content is also missing coordiantes. Based on that, i would actually be inclined to list them first

First also makes sense to me.

You can click on the test logs to see the detailed error, but here's the trace:

[richardye101]

I fixed the recursion error, was a mistake in copying the sorting code between grouped shapes and normal shapes