Child algorithm in custom Processing algorithm fails to find temporary output of previous child algorithm

Question

I wrote a simple custom QGIS Processing algorithm in Python:

from qgis.core import QgsProcessing
from qgis.processing import alg
import processing
@alg(name="test", label=alg.tr("test"), group="test", group_label=alg.tr("test"))
@alg.input(type=alg.SOURCE, types=[QgsProcessing.TypeVectorPoint], name="POINTS_INPUT", label="Point layer")
@alg.input(type=alg.RASTER_LAYER_DEST, name="OUTPUT", label="Raster output layer")
def test(instance, parameters, context, feedback, inputs):
 """Description goes here. (Don't delete this! Removing this comment will cause errors.)"""
 buffer = processing.run(
 "native:buffer",
 {
 "INPUT": parameters["POINTS_INPUT"],
 "DISTANCE": 100,
 "OUTPUT": "memory:",
 },
 is_child_algorithm=True,
 context=context,
 feedback=feedback,
 )
 
 feedback.pushInfo("Buffering done...")
 
 rasterize = processing.run(
 "gdal:rasterize",
 {
 'INPUT': buffer["OUTPUT"],
 'BURN': 123,
 'UNITS': 1,
 'WIDTH': 10,
 'HEIGHT': 10,
 'OUTPUT': QgsProcessing.TEMPORARY_OUTPUT,
 },
 is_child_algorithm=True,
 feedback=feedback,
 )
 return {"OUTPUT": rasterize["OUTPUT"]}

When I run it, the call to the gdal:rasterize algorithm fails to find its input though.

My QGIS is in German, so the error is German:

Kann Algorithmus nicht ausführen Konnte Quelllayer für INPUT nicht laden: output_ac05fad6_1b89_41ba_960e_a68be241e84b nicht gefunden

What is wrong?

Answer 1

processing.run() needs the context, otherwise it does not know where to find the temporary data. You should always pass the context and feedback objects to the child algorithms.

 rasterize = processing.run(
 "gdal:rasterize",
 {
 'INPUT': buffer["OUTPUT"],
 'BURN': 123,
 'UNITS': 1,
 'WIDTH': 10,
 'HEIGHT': 10,
 'OUTPUT': QgsProcessing.TEMPORARY_OUTPUT,
 },
 is_child_algorithm=True,
 context=context, # <- this line was added
 feedback=feedback,
 )