NumPyArrayToFeatureClass gives RuntimeError: create table

[GrzegorzJeleniewicz]

I've tried to change polygon feautre  to point by using tools:

FeatureClassToNumPyArray
NumPyArrayToFeatureClass.
My code as below:
import arcpy
import numpy as np
in_file='C:/ios/OneDrive/siatki_obliczeniowe/siatki.gdb/siatka_005'
a = arcpy.da.FeatureClassToNumPyArray(in_file, ["*"], explode_to_points=False)
out_file='C:/ios/OneDrive/siatki_obliczeniowe/siatki.gdb/xxxx3'

arcpy.da.NumPyArrayToFeatureClass(a, out_file,["shape"])
I got an error:

Traceback (most recent call last):
  File "C:/ios/OneDrive/python/feature_to_point.py", line 7, in <module>
    arcpy.da.NumPyArrayToFeatureClass(a, out_file,["shape"])
RuntimeError: create table


I had changed out_file to shp file and it works but it is not solution for me. I would like to be able write to .gdb.
Any idea how to solve this problem?

[DanPatterson_Retired]

your geometry field... what was it named? and what was it in the array?

you indicate it was ['shape']? or ['Shape@'] or 'SHAPE@' or some other variant

use

a.dtype‍‍

and get the exact case and spelling of the geometry field.

Also you will need to set

...explode_to_points=True)  # in order to get points, ‍‍

you can't set explode_to_points=False, unless you want the centroid of the original polygon

Thirdly, if the output featureclass name exists, it will fail

[GrzegorzJeleniewicz]

Thank you for your answer.

my geomtery field is ['Shape']. I've cheked it with a.dtype and got result:

[('OBJECTID', '<i4'), ('Shape', '<f8', (2,)), ('Shape_Length', '<f8'), ('Shape_Area', '<f8'), ('miasto', '<U50')]

I've changed small 's' for capital 'S', but it didn't help

Yes, I want the centroid of the original polygon, so explode_to_points=False is ok for me.

The output featureclass name is set to new each run.

[DanPatterson_Retired]

hmmm if you only want the centroid coordinate, you should avoid using the true shape, but use one of the alternates and specify your field names explictly

FeatureClassToNumPyArray—Data Access module | ArcGIS Desktop

you can easily parse the fields rather than using "*" which is ok in some situations, but the shape field should be parsed into x and y coordinates usually

flds = ['OID@', 'SHAPE@X', 'SHAPE@Y']

other_flds = [i.name for i in arcpy.ListFields(fc) if i.type not in ('OID', 'Geometry')]

flds.extend(other_flds)

flds

['OID@', 'SHAPE@X', 'SHAPE@Y', 'ID', 'Norm', 'Rank1', 'Text01', 'Text02', 'Xs', 'Ys',
 'Int_short', 'Float_fld', 'Grid_code', 'Rank_shp', 'PolyID', 'Grid_Count',
 'Diff_from_mean', 'x']‍‍‍‍‍‍‍‍‍‍‍

[GrzegorzJeleniewicz]

I changed my code following your instructions:

import arcpy
import numpy as np
in_file='C:/ios/OneDrive/siatki_obliczeniowe/siatki.gdb/strefy_miejskie_siatka_01'
flds = ['OID@', 'SHAPE@X', 'SHAPE@Y']
other_flds = [i.name for i in arcpy.ListFields(in_file) if i.type not in ('OID', 'Geometry')]
flds.extend(other_flds)
a = arcpy.da.FeatureClassToNumPyArray(in_file, flds, explode_to_points=False)
out_file='C:/ios/OneDrive/siatki_obliczeniowe/siatki.gdb/axxxyz9'
print (a.dtype)
arcpy.da.NumPyArrayToFeatureClass(a, out_file,['SHAPE@X', 'SHAPE@Y'])

Unfortunately I got error again:

Traceback (most recent call last):
  File "C:/ios/OneDrive/python/feature_to_point.py", line 10, in <module>
    arcpy.da.NumPyArrayToFeatureClass(a, out_file,['SHAPE@X', 'SHAPE@Y'])
RuntimeError: create table

The most annoying  fact is that if I change the 'out_file ' to shp file instead of feature class in geodatabase (.gdb) it works (with both code versions). For sure there is a problem with saving output to geodatabase (.gdb). I have no clue why.

[GrzegorzJeleniewicz · Accepted solution]

I got answer:

arcgis desktop - NumPyArrayToFeatureClass gives RuntimeError: create table - Geographic Information ... 

[DanPatterson_Retired]

Just noticed you are using python 2.7... may be the issue since I have no problem using the process with either points exploded or not exploded using python 3

[GrzegorzJeleniewicz]

I've got already the solution and posted link to it above.

[DanPatterson_Retired]

for future reference

all_flds = [f.name for f in arcpy.ListFields(fc, "*")]

editable_flds = [f.name for f in arcpy.ListFields(fc, "*") if f.editable]

all_flds

['OBJECTID', 'Shape', 'Grid_codes', 'F_', 'Shape_Length', 'Shape_Area', 'Rand_0',
 'Rand_1', 'Shape_Length_1', 'Shape_Area_1']

editable_flds

['Shape', 'Grid_codes', 'F_', 'Rand_0', 'Rand_1']‍‍‍‍‍‍‍‍‍‍‍‍

OBJECTID, and Shape info can be accessed through their @ property, other variants of Shape aren't editable and removing them from the list of candidates is a safe bet... at least for a file gdb