DM-27209: Update for new quantum graph API

[timj]

No description provided.

[natelust]

Some suggestions on how to do these operations more efficiently. What you wrote seems like it will work, so take whatever info you think is best.

[natelust]

quantum.inputs is already a mapping of DatasetType. As it is now, key is not used. If this was rearranged to:

...
for key, value in quantum_node.quantum.inputs.items():
    if key.name not in dataset_types_to_exclude:
        found = {butler.registiry.queryDatasets(key.name, collections=..., dataId=dataset.dataId) for dataId in value}
...

It would save a for loop over a lot of elements that will be excluded by dataset_types_to_exclude

[timj]

I have rejigged a bit based on the suggestion above but I can't use that line that does the loop over refs in queryDatasets because queryDatasets returns an interator and not DatasetRef and so you end up with a set of iterators. It's easier to use still use an explicit loop over refs (but still inside the if-check).

One thing that does confuse me is that the quantum_node.quantum.inputs is returning a proper DatasetRef with the id attached so the reality is that this queryDataset call is returning the same thing it had to start with. Not entirely sure that we expect that or not so maybe it's a special case for this particular example.

[natelust]

This would better be written as:

for quantum_node in graph.findQuantaWithDSType("raw"):
    if "raw" in quantum_node.quantum.inputs:  # Verify that raw is an input and not an output
        items.extend(quantum_node.quantum.inputs["raw"])

This should cut down on the number of iterations as findQuantaWithDSType uses precomputed dictionaries.