Editing Features¶
The ArcGIS API for Python makes programmatic editing of features a breeze. The edit_features() method on FeatureLayer object can be used for the same. In the example below, we see how to add, update and delete features from a point feature layer
#connect to GIS
from arcgis.gis import GIS
from IPython.display import display
gis = GIS("portal url", 'username', 'password')
#search for the feature layer named Ports along west coast
search_result = gis.content.search('title:Ports along west coast')
search_result[0]
Feature Layer Collection by arcgis_python_api
#access the item's feature layers
ports_item = search_result[0]
ports_layers = ports_item.layers
ports_layers
Query the features¶
Let us query the features on this layer
#query all the features and display it on a map
ports_fset = ports_layers[0].query() #an empty query string will return all
#the features or the first 1000 which ever is smaller
The df property of FeatureSet class is a powerful way to visualize all the features as a pandas dataframe table.
ports_fset.df
Check if the layer can be edited¶
To enable a feature layer to be edited, the edit capability needs to be turned on. This can be verified by accessing the capabilities property on the FeatureLayer object:
ports_flayer = ports_layers[0]
ports_flayer.properties.capabilities
Editing features¶
The feature layer does have all editing, updating and deleting capabilities turned on, thus it is ready for edit operations.
Updating features¶
From the dataframe above, we notice the short_form field of San Francisco port has a typo. We can update this by calling the edit_features() method on the ports_flayer FeatureLayer object. The edit_features() method accepts a Feature object with correct parameters set.
ports_features = ports_fset.features
# select San Francisco feature
sfo_feature = [f for f in ports_features if f.attributes['port_name']=='SAN FRANCISCO'][0]
sfo_feature.attributes
As we can see above, the short_form attribute should be changed to SFO. Let us edit that and store it as a new Feature object
sfo_edit = sfo_feature
sfo_edit.attributes['short_form'] = 'SFO'
display(sfo_edit)
Now let us send this feature object to the edit_features() method and edit the original feature layer
update_result = ports_flayer.edit_features(updates=[sfo_edit])
update_result
Adding features¶
From the data frame, we also notice that 'Los Angeles', an important port is missing. We can add this to the original layer by calling the edit_features() method and passing a dictionary representation of the feature
# construct a Feature object for Los Angeles.
la_dict = {"attributes":
{"latitude": 33.75,
"longitude": -118.25,
"country": "US",
"harborsize": "L",
"label_position": "SW",
"port_name": "LOS ANGELES",
"short_form": "LAX"},
"geometry":
{"x": -13044788.958999995, "y": 3857756.351200014}}
add_result = ports_flayer.edit_features(adds = [la_dict])
add_result
Deleting features¶
From the data frame, we notice 'Redlands', a land locked city is wrongly added as a port. We can remove it by calling the edit_features() method and passing the object id of the feature we want to be remove to the deletes parameter.
# find object id for Redlands
Redlands_feature = [f for f in ports_features if f.attributes['port_name'] == 'REDLANDS'][0]
Redlands_objid = Redlands_feature.get_value('objectid')
Redlands_objid
type(Redlands_objid)
# pass the object id as a string to the delete parameter
delete_result = ports_flayer.edit_features(deletes=str(Redlands_objid))
delete_result
Check results of edit operations¶
We have successfully added, removed and edited features. Let us query the layer once again and display the resutls as a dataframe and verify them.
ports_fset_edited = ports_flayer.query()
ports_fset_edited.df
Thus, we no longer see 'Redlands' which we deleted. We see a new objectid of 23 corresponding to Los Angeles that we added. Also the short_form of San Francisco is corrected to 'SFO'.
The edits we performed on the features are persisted in the feature layer and will reflect in all clients accessing it.
Feedback on this topic?