vis.show() UnicodeEncodeError

[hutaohutc]

I just run the tutorial code

import warnings
from causalnex.structure import StructureModel

warnings.filterwarnings("ignore")  # silence warnings

sm = StructureModel()
sm.add_edges_from([ ('health', 'absences'), ('health', 'G1') ])
from causalnex.plots import plot_structure, NODE_STYLE, EDGE_STYLE

viz = plot_structure(
    sm,
    all_node_attributes=NODE_STYLE.WEAK,
    all_edge_attributes=EDGE_STYLE.WEAK,
)
viz.show("01_simple_plot.html")

but get error: UnicodeEncodeError: 'gbk' codec can't encode character '\xa9' in position 230710: illegal multibyte sequence, I have changed 2 win PCs,but get the same error

[kylelim-mckinsey]

Hi! this can be solved if you include the following code

import os
os.environ["PYTHONIOENCODING"] = "utf-8"

The error message UnicodeEncodeError: 'gbk' codec can't encode character '\xa9' in position 230710: illegal multibyte sequence suggests that you're encountering an issue with character encoding. It appears that your Python environment is trying to use the GBK codec, which is typically used for Chinese characters, but it's encountering a character (in this case, '\xa9') that it cannot encode.

However, in your code, you're dealing with the graph visualization which shouldn't cause such error. So, it's likely this issue is environment-related rather than directly related to your code.

One workaround is to set the Python environment to use UTF-8 encoding by default. You can achieve this by setting the environment variables in Python:

[hutaohutc]

@kylelim-mckinsey thank you for your answer,but I use this code still get error.

UnicodeEncodeError: 'gbk' codec can't encode character '\xa9' in position 230710: illegal multibyte sequence

[kylelim-mckinsey]

Let's use this. (I forgot to put the hyphen. It is 'utf-8'

import os
os.environ['PYTHONIOENCODING'] = 'utf-8'

So based on the error message, you are likely experiencing an encoding issue while using the Graphviz library in Python. This is typically due to differences in default encodings across systems. As the error message suggests, the issue might be related to the 'gbk' codec not being able to encode a specific character. 'gbk' is an encoding primarily used for simplified Chinese characters.

When Python interacts with your operating system or file system, it uses the system's default encoding. In Windows, it's often a variant of 'gbk'. It seems like the function you're using (which indirectly relies on Graphviz) is trying to write a character not compatible with the 'gbk' encoding.

If that doesn't work, another workaround might be to use the pygraphviz library directly to generate the graph, since this library has a draw method.

[hutaohutc]

I am sorry ,I still get the same error..

[GabrielAzevedoFerreiraQB]

I see. Thanks for the issue, Tau. It seems the setting of encoding is somehow not taking effect.

This is a pyvis issue which we reported and which seems to be related to your specific Windows encoding. This is a common issue and we raised a PR to fix on next releases (https://github.com/WestHealth/pyvis/pull/231)

In the meantime we suggest:

• changing the default encoding to utf-8 (what we suggested is somehow not be taking effect) or
• not using the show method and instead do the show steps manually (with the right encoding). Below, I am mimicking the steps from https://github.com/WestHealth/pyvis/blob/master/pyvis/network.py#L537 - Just as a pseudo code

# define width, height and name for file
name = "index.html"
height="600px"
width="100%"

# generate html
html = viz.generate_html(notebook=notebook)

# write html
with open(name, "w+", encoding="utf-8") as out:
    out.write(html)

#Display as an IFrame
IFrame(name, width=width, height=height)

I will close the issue, but let us know if you encounter any issues. I am happy to reopen it again

[wxq0309]

I just ran the simplest demo, how can I solve this problem？

[GabrielAzevedoFerreiraQB]

Could you try the instructions above, please?

as explained above this is an issue with PyVis (a different package) in your specific setup (message above has more details)

to get around the issue, you should not use viz.show() but instead use the longer command above.

This is because pyvis unfortunately expects an utf-8 encoding, which is not your set up seems to be.

[hutaohutc]

@GabrielAzevedoFerreiraQB It's worked! Thanks .

[GabrielAzevedoFerreiraQB]

Awesome news. thanks!