Python, unable to import Pandas in Python3Component

Scripting
, , ,
1

Hello,

Like many other colleagues, we are still facing the error : "1. Error running script: partially initialized module ‘pandas’ has no attribute ‘_pandas_datetime_CAPI’ (most likely due to a circular import) [40:1]" when trying to import pandas.

Are there any stable solutions to come? Does anyone knows if this problem is at Rhino’s side?

It seems very hard to fix.

Thank you,

LS

3

Hi, I’m not sure about the technical explanation, but this method is very effective for me. You can check out this link: Gmsh in Grasshopper - #6 by Rh-3d-p

4

Hi, thanks for your help :slight_smile: Unfortunately it seems this didnt work for pandas explicitly. It did the job for numpy tho.

So my question remains: What is going on with the Pandas package?

image
image689×809 20.8 KB

Also tried with pandas 2.0 instead of 2.2 but I get the locale problem. I do think this problem is on rhino’s side. Can someone confirm?

image
image634×780 35.1 KB

1 Like
5

This fixed it for now:

First I downgraded pandas to 2.0

pip install --upgrade --force-reinstall pandas==2.0
  • Then i set the regions to English (united states):
  1. Go to Settings > Time & Language > Language & Region.
  2. Check if English (United States) is available under the Languages section. If it’s not, add it by clicking on “Add a language” and searching for English (United States).
  3. Under Region, ensure that the United States is selected.
  4. Open Control Panel.
  5. Go to Clock and Region > Region.
  6. Click on the Administrative tab.
  7. Click on the Change system locale button.
  8. Make sure English (United States) is selected, and then click OK.

and then I set manually the locale in the Rhino Python script:

import locale
locale.setlocale(locale.LC_ALL, 'en_US.UTF-8')

import pandas as pd

image
image1356×889 90.6 KB

Hope this might be usefull for anyone.

Leandre Sassi

3 Likes
7

Followed the solution suggested by @leandre.sassi which did indeed fix the _pandas_datetime_CAPI issue, however raising another with numpy.dtype:

1. Error running script: numpy.dtype size changed, may indicate binary incompatibility. Expected 96 from C header, got 88 from PyObject [7:1]

Anyone with an idea what the cause is, and how it may be solved?

8

The numpy.dtype error seems to have occurred as pandas 2.0 was compiled against an earlier version of numpy and doesn’t support numpy 2.0. Fixed by also downgrading numpy to some version < 2.0.

# requirement pandas==2.0
# requirement numpy==1.26.4