Passing Geometry object to processing tool instead of Layer

Question

In ArcPy, you can pass a geometry object as parameter to geoprocessing tool:

import arcpy
coordinates = [[20, 30], [30, 30], [30, 20], [20, 20]]
array = arcpy.Array([arcpy.Point(x, y) for x, y in coordinates])
Create a polygon geometry object using the array object
boundary = arcpy.Polygon(array, arcpy.SpatialReference(4326))
arcpy.Clip_analysis('c:/data/rivers.shp', 
                     boundary, # <- GEOMETRY OBJECT
                    'c:/data/rivers_clipped.shp')

I have a polygon geometry and a layer. I want to clip the layer by the polygon geometry using PyQGIS. Normally, I make a new layer, add the geometry to the layer and use it in processing.run as follows:

coordinates = [QgsPointXY(20, 30), QgsPointXY(30, 30),
               QgsPointXY(30, 20), QgsPointXY(20, 20)]
geometry = QgsGeometry.fromPolygonXY([coordinates ])
temp_layer = QgsVectorLayer(f"Polygon?crs=EPSG:4326", "TEST", "memory")
feature = QgsFeature()
feature.setGeometry(geometry)
temp_layer.dataProvider().addFeatures([feature])
temp_layer.updateExtents()
LayerA = iface.activeLayer()
result = processing.run("native:difference",
                        {'INPUT': LayerA,       # <- QgsVectorLayer
                         'OVERLAY': temp_layer, # <- QgsVectorLayer
                         'OUTPUT':'TEMPORARY_OUTPUT'})

But I would like to avoid making a new layer for one geometry. I tried passing a geometry to processing.run, but it obviously didn't work:

result = processing.run("native:difference",
                        {'INPUT': LayerA,       # <- QgsVectorLayer
                         'OVERLAY': geometry,   # <- QgsGeometry ?
                         'OUTPUT':'TEMPORARY_OUTPUT'})

How can I pass a geometry object to a processing tool without making a new layer for the geometry? Is that possible in PyQGIS?

This something I have wondered too. Currently I have a function, geom2lyr to convert a geometry or list of geometries into a layer, in a library that I import into most of my scripts. Using the geometry directly would be very handy. — Matt, Jan 25 '22 at 14:18

Taras · Answer 1 · 2022-05-12T08:06:03.380

When I check the help for the "native:difference"/"qgis:difference" algorithm with processing.algorithmHelp("native:difference") it declares that:

INPUT: Input layer
Parameter type: QgsProcessingParameterFeatureSource

Accepted data types:
    - str: layer ID
    - str: layer name
    - str: layer source
    - QgsProcessingFeatureSourceDefinition
    - QgsProperty
    - QgsVectorLayer

where:

QgsProcessingFeatureSourceDefinition : a constructor for QgsProcessingFeatureSourceDefinition, accepting a static string source.
QgsProperty : a constructor for a QgsAbstractProperty. The property will be set to an InvalidProperty type.
QgsVectorLayer : represents a vector layer which manages a vector based data sets.

So, it seems there is no possibility to pass directly the geometry object to processing algorithm.

However, I can suggest an alternative that uses a "Virtual Layer":

coordinates1 = [QgsPointXY(20, 30), QgsPointXY(30, 30), QgsPointXY(30, 20), QgsPointXY(20, 20)]
coordinates2 = [QgsPointXY(25, 35), QgsPointXY(35, 35), QgsPointXY(35, 25), QgsPointXY(25, 25)]
geometry1 = QgsGeometry.fromPolygonXY([coordinates1])
geometry2 = QgsGeometry.fromPolygonXY([coordinates2])
layer1 = QgsVectorLayer( f"?query=SELECT ST_GeomFromText('{geometry1.asWkt()}')", "vlayer1", "virtual" )
layer2 = QgsVectorLayer( f"?query=SELECT ST_GeomFromText('{geometry2.asWkt()}')", "vlayer2", "virtual" )
result = processing.runAndLoadResults("native:difference",
                        {'INPUT': layer1, # <- QgsVectorLayer
                         'OVERLAY': layer2, # <- QgsVectorLayer
                         'OUTPUT':'memory:'})

If it is not mandatory to use the "Difference" tool one can also try this:

coordinates1 = [QgsPointXY(20, 30), QgsPointXY(30, 30), QgsPointXY(30, 20), QgsPointXY(20, 20)]
coordinates2 = [QgsPointXY(25, 35), QgsPointXY(35, 35), QgsPointXY(35, 25), QgsPointXY(25, 25)]
geometry1 = QgsGeometry.fromPolygonXY([coordinates1])
geometry2 = QgsGeometry.fromPolygonXY([coordinates2])
difference_geometry = geometry1.difference(geometry2)
result = QgsVectorLayer( f"?query=SELECT ST_GeomFromText('{difference_geometry.asWkt()}')", "result", "virtual" )
QgsProject.instance().addMapLayer(result)

References:

Passing Geometry object to processing tool instead of Layer

Create a polygon geometry object using the array object

1 Answers1

Linked