Exporting an image

[harrylojames]

How would I export a plot to html/png/svg?

Ideally I would like to be able to do something similar to Plotly i.e. fig.write_html("path/to/file.html") / fig.write_image("images/fig1.png") / fig.write_image("images/fig1.svg")

[juba]

Hi, this is a good question.

I don't think there is a simple way to do this right now. I should check how to implement this, but it could take some time.

[harrylojames]

Thank you for giving this thought! Any efforts would be greatly appreciated as it's a significant blocker to adopting this package in my workflows more generally

[juba]

Hi, and sorry for taking this long for looking into this issue.

I added a path argument to plot specification when using a jsdom renderer to allow to output the plot to an HTML or SVG file. So you should be able to do something like this:

from pyobsplot import Obsplot, Plot

oj = Obsplot(renderer="jsdom", theme="light")
oj(Plot.rectY(range(3), Plot.binX({"y": "count"})), path="out.html")

This is still experimental and only available in the github version.

Implementing this for the widget renderer doesn't work for the moment. But I believe that even if it works one day, it will only be able to generate a non-easily embeddable HTML file. But I'll continue to take a look into it.

[harrylojames]

No stress at all. Things take time and when I briefly searched online this looked like it would be a tough nut to crack!

Thanks so much for knocking something up. I'll give it a go.

[juba]

In fact I just published pyobsplot 0.4.0 which allows to output plots generated by the jsdom renderer to an SVG or HTML file.

There is a bit of documentation about it here:

https://juba.github.io/pyobsplot/usage.html#saving-plots-to-file

Any feedback appreciated !

Output to file doesn't work for the widget renderer, but I'm not sure if it really makes sense anyway.

[harrylojames]

Hi thanks for publishing!

Using the latest version with python 3.9 gives the following error.

     45     self.port = port
     46     self.theme = theme
---> 48 def plot(self) -> SVG | HTML:
     49     """Generates the plot by sending request to http node server.
     50 
     51     Returns:
     52         Either an HTML or SVG IPython.display object.
     53     """
     55     # Make POST request with plot spec

TypeError: unsupported operand type(s) for |: 'type' and 'type'

The | operator was added in Python 3.10. Given 3.12 was released this week it might be reasonable to set the minimum Python version as 3.10 but your call!

Not a blocker for me as I'm happy to update my Python version :)

[juba]

Ah, you're right, thanks for pointing this out. I found this bug earlier but only put a workaround for CI tests.

The development version should be compatible with Python 3.9 again, sorry !

[juba]

Thanks to the solution in manzt/anywidget#339, saving widget output to an HTML file should now be possible. I'll try to publish a new package version shortly.

[juba]

v0.4.1 just released: https://github.com/juba/pyobsplot/releases/tag/v0.4.1

[harrylojames]

Amazing! Thank you so much for looking into this. I will give it a go tomorrow and let you know how it goes

[harrylojames]

PR to address a bug with saving SVGs #22

[harrylojames]

Another update that I hope is helpful.

When exporting HTML the title font from my notebook is not preserved.

Perhaps this is an observable plot question and not a pyobsplot question!

op = Obsplot(renderer="jsdom")

op(
    {
        "marks": [
            Plot.tickX([0, 5, 10, 15])
        ], 
        "title" : "a",
        "style":{"font-family":"serif"},
    },
    path="test.html"
)

Do you know if this is fixable? Including font-family changes the font for the axis elements but not the title or caption.

I noticed Plotly preserves the title element when saving to SVG with Kaleido

[juba]

When you display the plot without exporting it to a file, your title is in a serif font ? As far as I can see the style option is applied to the <svg> element in the output, so it shouldn't change the font of title, subtitle, caption or legend.