Update Definition on Layer in Feature Service

Question

Have been tinkering with this for the better part of a day and can't figure it out. I'm calling an API to get a json response containing store locations, which contain the coordinates, then pushing that into a Feature Layer. Esri's implementation of the Spatially Enabled Dataframe's .to_featurelayer has the unfortunate behavior of sanitizing column names, even if the column names are valid to begin with. The result is that its changing my column names from their original case to snake_case. For example 'storeName' gets changed to 'store_name' when I utilize the .to_featurelayer method to publish the data frame as a feature layer, even though there is absolutely nothing invalid at all about a feature layer with a field named 'storeName'. Unlike the .to_featureclass method, where sanitize_columns is exposed as a parameter in the method and defaulted to True, meaning you can set it to False to avoid this behavior, sanitize_columns is defaulted to to True in the .to_featurelayer method and is not exposed as a parameter so there is no way to avoid it. As a result, I'm trying to go back and update the column names using the update_definition method on the feature layer manager but I keep getting one of two errors. The first approach I took: from arcgis.features import FeatureLayer featureLayer =gis.content.get("c952e9e257bd4fc887be2934291548cb") lyr = featureLayer.layers[0] lyr.properties originalDefinition = lyr.properties # Get the fields array originalFields = originalDefinition["fields"] newFields = originalFields.copy() for f in newFields: if "_" in f["name"]: newFieldName = f["name"].split("_")[0].lower()+f["name"].split("_")[1].title() f["name"] = newFieldName #newFields print('"fields": ' + json.dumps(newFields)) lyr.manager.update_definition('"fields": ' + json.dumps(newFields)) Fails with: --------------------------------------------------------------------------- Exception Traceback (most recent call last) in 13 #newFields 14 print('"fields": ' + json.dumps(newFields)) ---> 15 lyr.manager.update_definition('"fields": ' + json.dumps(newFields)) /opt/conda/lib/python3.7/site-packages/arcgis/features/managers.py in update_definition(self, json_dict) 2002 u_url = self._url + "/updateDefinition" 2003 -> 2004 res = self._con.post(u_url, params) 2005 self.refresh() 2006 return res /opt/conda/lib/python3.7/site-packages/arcgis/gis/_impl/_con/_connection.py in post(self, path, params, files, **kwargs) 718 file_name=file_name, 719 try_json=try_json, --> 720 force_bytes=kwargs.pop('force_bytes', False)) 721 #---------------------------------------------------------------------- 722 def put(self, url, params=None, files=None, **kwargs): /opt/conda/lib/python3.7/site-packages/arcgis/gis/_impl/_con/_connection.py in _handle_response(self, resp, file_name, out_path, try_json, force_bytes) 512 return data 513 errorcode = data['error']['code'] if 'code' in data['error'] else 0 --> 514 self._handle_json_error(data['error'], errorcode) 515 return data 516 else: /opt/conda/lib/python3.7/site-packages/arcgis/gis/_impl/_con/_connection.py in _handle_json_error(self, error, errorcode) 534 535 errormessage = errormessage + " (Error Code: " + str(errorcode) +")" --> 536 raise Exception(errormessage) 537 #---------------------------------------------------------------------- 538 def post(self, Exception: Unable to update feature service layer definition. Object reference not set to an instance of an object. (Error Code: 400) Kinda makes sense. Maybe I need to instantiate a FeatureLayer object on the actual layer...Let's try that: featureLayer =gis.content.get("c952e9e257bd4fc887be2934291548cb") lyr = FeatureLayer(featureLayer.layers[0]) originalDefinition = lyr.properties # Get the fields array originalFields = originalDefinition["fields"] newFields = originalFields.copy() for f in newFields: if "_" in f["name"]: newFieldName = f["name"].split("_")[0].lower()+f["name"].split("_")[1].title() f["name"] = newFieldName newFields print('"fields": ' + json.dumps(newFields)) #lyr.manager.update_definition('"fields": ' + json.dumps(newFields)) But it fails with: --------------------------------------------------------------------------- Exception Traceback (most recent call last) /opt/conda/lib/python3.7/site-packages/arcgis/gis/__init__.py in _hydrate(self) 11481 if isinstance(self._con, Connection): > 11482 self._lazy_token = self._con.generate_portal_server_token(serverUrl=self._url) 11483 else: /opt/conda/lib/python3.7/site-packages/arcgis/gis/_impl/_con/_connection.py in generate_portal_server_token(self, serverUrl, expiration) 1313 resp = self.post(path=self._token_url, postdata=postdata, -> 1314 ssl=True, add_token=False) 1315 if isinstance(resp, dict) and resp: /opt/conda/lib/python3.7/site-packages/arcgis/gis/_impl/_con/_connection.py in post(self, path, params, files, **kwargs) 719 try_json=try_json, --> 720 force_bytes=kwargs.pop('force_bytes', False)) 721 #---------------------------------------------------------------------- /opt/conda/lib/python3.7/site-packages/arcgis/gis/_impl/_con/_connection.py in _handle_response(self, resp, file_name, out_path, try_json, force_bytes) 513 errorcode = data['error']['code'] if 'code' in data['error'] else 0 --> 514 self._handle_json_error(data['error'], errorcode) 515 return data /opt/conda/lib/python3.7/site-packages/arcgis/gis/_impl/_con/_connection.py in _handle_json_error(self, error, errorcode) 535 errormessage = errormessage + " (Error Code: " + str(errorcode) +")" --> 536 raise Exception(errormessage) 537 #---------------------------------------------------------------------- Exception: Unable to generate token. 'username' must be specified. 'password' must be specified. 'referer' must be specified. (Error Code: 400) During handling of the above exception, another exception occurred: AttributeError Traceback (most recent call last) in 2 featureLayer =gis.content.get("c952e9e257bd4fc887be2934291548cb") 3 lyr = FeatureLayer(featureLayer.layers[0], gis=gis) ----> 4 lyr.properties 5 originalDefinition = lyr.properties 6 # Get the fields array /opt/conda/lib/python3.7/site-packages/arcgis/gis/__init__.py in properties(self) 11460 return self._lazy_properties 11461 else: > 11462 self._hydrate() 11463 return self._lazy_properties 11464 /opt/conda/lib/python3.7/site-packages/arcgis/gis/__init__.py in _hydrate(self) 11507 # try as a public server 11508 self._lazy_token = None > 11509 self._refresh() 11510 11511 except HTTPError as httperror: /opt/conda/lib/python3.7/site-packages/arcgis/gis/__init__.py in _refresh(self) 11450 dictdata = self._con.get(self.url, params) 11451 else: > 11452 raise e 11453 11454 self._lazy_properties = PropertyMap(dictdata) /opt/conda/lib/python3.7/site-packages/arcgis/gis/__init__.py in _refresh(self) 11443 else: 11444 try: > 11445 dictdata = self._con.post(self.url, params, token=self._lazy_token) 11446 except Exception as e: 11447 if hasattr(e, 'msg') and e.msg == "Method Not Allowed": /opt/conda/lib/python3.7/site-packages/arcgis/gis/_impl/_con/_connection.py in post(self, path, params, files, **kwargs) 619 try_json = kwargs.pop("try_json", True) 620 add_token = kwargs.pop('add_token', True) --> 621 if url.find('://') == -1: 622 url = self._baseurl + url 623 if kwargs.pop("ssl", False) or self._all_ssl: AttributeError: 'FeatureLayer' object has no attribute 'find' I don't understand why I would need to authenticate with credentials manually here or even how I would. I have a connection to a GIS object and even passing that in as the gis parameter on the instantiation of the FeatureLayer object at line 3 doesn't change result. I also tried using FeatureCollection but got the same result in both approaches. Help or insight appreciated.

AndrewChapkowski · Accepted Answer

You cannot update the field names directly, you need to use a combination of update definition, add to definition and delete from definition to rename columns.

Let's start with the update call.

update_def = {
    "fields": [
        {
            "name": "THISisWrong",
            "type": "esriFieldTypeString",
            "alias": "AmazingAlias",
            "length": 256,
            "editable": True,
            "nullable": True,
            "defaultValue": None,
            "description": None,
            "domain": None,
        }
    ]
}
res = flyer.manager.update_definition(update_def)
print(res)

This will update the alias column, though it doesn't directly answer your question.

Now how can we take say column: ThisIsWrong and Rename it to this_is_wrong? or something like that.  You'll have to do it in a few steps:

* Add the new field
* Copy the column data* If field is uneditable, make it uneditable

* drop the old field

import re
from arcgis.gis import GIS
from arcgis.features import FeatureLayer

def _camelCase_to_underscore(name):
    """PEP8ify name"""
    if name[0].isdigit():
        name = "execute_" + name
    name = name.replace(" ", "_")
    if '_' in name:
        return name.lower()
    s1 = re.sub('(.)([A-Z][a-z]+)', r'\1_\2', name)
    return re.sub('([a-z0-9])([A-Z])', r'\1_\2', s1).lower()

if __name__ == "__main__":

gis = GIS(profile='your_online_profile', verify_cert=False, trust_env=True)
    item = gis.content.get("e0d177f6b2ed4b1db1abe848dfea23f8")
    flyer = item.layers[0]

#
    # Creating Columns with New Names
    #
    unchange_fields = [
        v
        for k, v in dict(flyer.properties).items()
        if k in ['objectIdField', 'globalIdField']
    ]
    rename_fields = [
        dict(fld)
        for fld in flyer.properties.fields
        if fld["type"] not in ('esriFieldTypeOID', 'esriFieldTypeGlobalID')
        or fld['name'] not in (unchange_fields)
    ]
    expressions = []
    delete_exp = {"fields": []}
    add_fields = []
    uneditable = []
    for idx, fld in enumerate(rename_fields):
        if _camelCase_to_underscore(fld["name"]) != fld["name"].lower():

expressions.append(
                {
                    "field": _camelCase_to_underscore(fld["name"]),
                    "sqlExpression": fld["name"],
                }
            )
            delete_exp["fields"].append({"name": fld["name"]})

fld['name'] = _camelCase_to_underscore(fld["name"])
            fld['alias'] = _camelCase_to_underscore(fld["alias"])
            if fld['editable'] == False:
                uneditable.append(fld['name'])
            fld['editable'] = True
            add_fields.append(fld)
        else:
            print(f"Skipping field because it meets the format standards: {fld['name']}")

print("Add the renamed columns")

res = flyer.manager.add_to_definition({"fields": add_fields})
    print(res)
    print("Update the column values")
    job = flyer.calculate(where="1=1", calc_expression=expressions, future=True)
    result = job.result()
    print(result)

# Set fields that are not editable back to original state.
    #
    update_editable = [fld for fld in add_fields if fld['name'] in uneditable]
    if update_editable:
        for fld in update_editable:
            fld['editable'] = True
        update_def = {"fields": update_editable}

res = flyer.manager.update_definition(update_def)
        print(res)
    #  Drop the Old Fields
    #
    print(flyer.manager.delete_from_definition(delete_exp))
    flyer._refresh()
    print([fld['name'] for fld in flyer.properties.fields])

It should be noted that this code doesn't handle cases.  If you have a field called Editor the caml casing should be editor.  The underlying database sees the field as the same name and thus can't be renamed.