If you're on macOS, definitely check out Monodraw [1] as a quick and easy way to create such diagrams (then copy + paste into ShakyDraw). Just remember to set your document alphabet to ASCII and you're good to go (from the inspector).
Full disclosure: I'm the developer of Monodraw. If you have any questions, happy to answer here.
That's a very good question that comes up quite often. Unfortunately, that's not currently supported :(
In fact, it's technically impossible to recover all the information that was originally present in the Monodraw document because the conversion to text is inherently lossy. Now, you can probably try to use some heuristics to recover as much information as possible (e.g., detect boxes, arrows). That would certainly work in many simple cases where there's no overlapping shapes.
Having said that, I've got other, higher priority updates planned in the pipeline before I can even consider whether to tackle this problem.
I want the opposite. Draw a block diagram on paper, scan it, and get a super swank omnigraffle-style diagram from it (or a top-notch ASCII art diagram).
This looks great. I can't say the same about their website which feels like a mess and really doesn't sell what otherwise seems like a very useful app.
Came here wanting this, found it in the comments. OP, it's cool and all - but I'm never going to manually type out box diagrams. I don't write e-zines or NFO's for warez releases.
While going the other way is not possible directly, but you can use something like asciiflow.com to make a ASCII diagram easily and paste it in ShakyDraw.
That's exactly what I was going to say - the "hand drawn" diagram on the right is better looking, but the left hand is far more accessible and arguably have more utility.
MyScript offers an API that does this. You need to feed it all the motions/sequence of points that make up the shape (known as "online OCR"), it won't work with a static document.
If you like this, try SvgBobRus[0], which also converts plain-text diagrams into graphics (and has a larger vocabulary). There's also Spongedown[1], which makes it possible to embed these diagrams in your Markdown files and render the whole thing to HTML + SVG.
And where is stated mraleph codebase to go from ascii to an image? I'd like a command line tool to convert locally, or this open sourced so I can host a local version.
This looks like it uses a very similar algorithm to https://github.com/dbushong/shaky , which is a port of @mraleph's 2012 talk (@mraleph is cited in this site's about page)
Aw. I wrote something last year to translate old documents with plain text boxes and arrows to Unicode line drawing characters. This was a small part of another project. It's here:
If it had that, somebody would wish you could draw arrows, move and resize boxes by dragging, etc. Then you've got a full blown ASCII drawing program. They already exist as a few people have mentioned in this thread. So you can use them for what they're good at and copy/paste their output into ShakeyDraw.
I thought that at first - but what if you want a digital copy; what's the plainest, whitest thing you can find to draw on, and do you have a scanner?
Obviously a lot of the time if you're going to the deliberate trouble of creating a digital image, you'd probably prefer it didn't look hand drawn... but I can imagine seeing it in repo readmes; that sort of thing. Actually that's quite appealing - because you could source control the ASCII too.
Full disclosure: I'm the developer of Monodraw. If you have any questions, happy to answer here.
[1] https://monodraw.helftone.com