System Error when exporting Spatially Enabled DataFrame to Feature Class

[DarrenConly]

Hi All,

I'm trying to do the following:

1. Convert a non-spatial pandas dataframe to a spatial dataframe
2. Export the spatial dataframe to a feature class in an ESRI file geodatabase, based on the instructions provided in ESRI's Introduction to the Spatially Enabled DataFrame

However, I get the error "SystemError: <da.funcInfo object at 0x0000029FA43C52A0> returned NULL without setting an error".

I googled the error and the only hits I got implied the issue has something to do with making a C extension. Do any of you have any insights on this?

Here's the full code for context:

out_fc_path = 'I:\\Projects\\Darren\\EmpInventory\\EmploymentInventory.gdb\\testrecs9581720210407_1435'

sdft = GeoAccessor.from_xy(master_df, fld_lat, fld_lon) # no issues here!

sdft.spatial.to_featureclass(location=out_fc_path)
Traceback (most recent call last):

  File "<ipython-input-29-2e5a93021399>", line 1, in <module>
    sdft.spatial.to_featureclass(location=out_fc_path)

  File "C:\Users\dconly\AppData\Local\ESRI\conda\envs\arcgispro-py3-dcmar21\lib\site-packages\arcgis\features\geo\_accessor.py", line 2122, in to_featureclass
    has_m=has_m)

  File "C:\Users\dconly\AppData\Local\ESRI\conda\envs\arcgispro-py3-dcmar21\lib\site-packages\arcgis\features\geo\_io\fileops.py", line 708, in to_featureclass
    arcpy.da.ExtendTable(fc, oidfld, array, join_dummy, append_only=False)

SystemError: <da.funcInfo object at 0x0000029FA43C52A0> returned NULL without setting an error

[DarrenConly · Accepted solution]

I found the solution!

In trying to find a workaround, I discovered that you cannot export a spatial dataframe to a file or feature class if it contains categorical data types. To fix this, you must convert all categorical data types to string data type.

Example: if you have a column 'X' that is categorical, you must first convert to string data type by doing the following:

df['X'] = df['X'].astype('str')

I'm happy to have resolved this, and hope my solution is informative to others.

Cheers,

[nzjs]

Darren, this led me in the right direction to solve similar issue. In my case I had a number of seemingly invalid data types, so this small function fixed it.

Cheers, John

def convert_dtypes_arcgis(df):
  # Convert dataframe dtypes which are not compatible with ArcGIS
  # Use builtin Pandas dtype conversion
  df = df.convert_dtypes(infer_objects=True)
  # Then str convert any remaining special object/category fields 
  for col in df.columns:
    # print(col, '/', df[col].dtype)
    if df[col].dtype == 'object' or df[col].dtype == 'category':
      df[col] = df[col].astype('str')
  # Return modified df
  return df

[MinbinJiang]

Thanks for posting the solution

[feralcatcolonist]

@MinbinJiang is this going to become a built-in for the API? I recently ran into this issue and am glad there's a workaround, but this seems to be a pretty significant defect.