Feature Request: Apply this to networkx pos layout argument

[jolespin]

This is totally off topic but I feel like your algorithm would be perfect for fixing networkx plots. Would be possible to adapt your code to take in a dictionary like this: {'node_A':[x,y], 'node_B':[x,y], ..., 'node_Z':[x,y]} w/ labels and then output a new node dictionary so when you plot the labels there's no overlap :)

This could be EXTREMELY useful and your code pretty much already solves the problem...

[Phlya]

Interesting. I have never used networkx, so not sure I fully understand what you are suggesting - could you show an example?

A problem is that, it seems, you want to provide strings which will be plotted by adjust_text, but it needs to have Text objects to start with - and while it could create them, not sure this is the right approach, since then it would have to expose all the options for text formatting, and that's a lot of stuff...

[jolespin]

Here’s on example of using one of the layout algorithms and some of the problems posed by it https://stackoverflow.com/q/14283341/678572 Yea, this is definitely out of the scope of adjusttext but your implementation just works so damn well that I thought I would bring it to your attention that you could definitely apply it to this problem . Those network layouts never consider text or labels.

…

On Jun 22, 2019, at 3:29 AM, Ilya Flyamer ***@***.***> wrote: Interesting. I have never used networkx, so not sure I fully understand what you are suggesting - could you show an example? A problem is that, it seems, you want to provide strings which will be plotted by adjust_text, but it needs to have Text objects to start with - and while it could create them, not sure this is the right approach, since then it would have to expose all the options for text formatting, and that's a lot of stuff... — You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub, or mute the thread.

[Phlya]

Would you say outputting final coordinates of the text objects would be sufficient to add the graph to the plot after adjusting text positions? This way you could create text objects according to original graph coordinates, adjust them, then change the graph to plot it so it fits the label coordinates? As I said, I am wary of implementing the original proposed idea, but happy to help to create a way to make it work.

[jolespin]

Yea, I think that would be the most preferred output if it was just adjusted coordinates that take into consideration the original positions and the text label size. Thanks for considering this.

…

On Jun 23, 2019, at 5:53 AM, Ilya Flyamer ***@***.***> wrote: Would you say outputting final coordinates of the text objects would be sufficient to add the graph to the plot after adjusting text positions? This way you could create text objects according to original graph coordinates, adjust them, then change the graph to plot it so it fits the label coordinates? As I said, I am wary of implementing the original proposed idea, but happy to help to create a way to make it work. — You are receiving this because you authored the thread. Reply to this email directly, view it on GitHub, or mute the thread.