Hex (encoded images) in `\pict` control groups is not removed. about pyth HOT 5 OPEN

Alright, so this is from this pull request, which I perhaps didn't think hard enough about: https://github.com/brendonh/pyth/pull/19/files

The stated purpose of that PR is to make it easier to filter out image data, by identifying it as such in the document object. But then none of the writers actually filter it out. Sigh.

I figure the right fix is to make Image a top-level type (instead of a Paragraph subclass), and then update the writers to ignore it. Or something.

from pyth.

Having looked again, it looks like the images are getting recognized. I see image objects inside of paragraphs, so it may be as easy just having the writers ignore it.

from pyth.

Right. So there are two bugs:

1. Image is a Paragraph subclass, leading writers to interpret it as text instead of (correctly) crashing on an unknown type
2. Writers don't know to skip it.

Someone should fix that! ;-)

from pyth.

@brendonh You should have highlighted me, since I wrote that original image support... I am not sure making an Image a top level class is the correct approach, since images actually appear in the flow of the paragraphs, and currently pyth checks religiously that a Document only contains a Paragraph.

Before my patch #19 Image data would just be interpreted as plain text, and so the output of all the writers hasn't changed. I currently think adding functionality to the writers to ignore/handle the images is the best way forward.

from pyth.

Ugh you are right on both counts. Okay, I'll do something about it.

from pyth.