after_pipeline_run hook run_result is empty on successful run

[Dekermanjian]

Description

The run_result parameter is an empty dictionary when the pipeline runs and saves the output successfully.

Context

I am trying to get the pipeline outputs in a hook

Steps to Reproduce

in hooks.py:

    @hook_impl
    def after_pipeline_run(self, run_params, run_result, pipeline, catalog) -> None:
        logger = logging.getLogger(__name__)

        if run_params["pipeline_name"] == "data_processing":
            logger.info(f"run_params: {run_params}")
            logger.info(f"run_result: {run_result}")
            logger.info(f"pipeline: {pipeline}")
            logger.info(f"catalog: {catalog}")

Expected Result

run_result contains the pipeline output

Actual Result

run_result = {}

Your Environment

kedro version 0.19.10
python version 3.12.7
linux

[ankatiyar]

Hi @Dekermanjian,
The runner returns the "free outputs" i.e. MemoryDatasets that are not registered in the catalog. If there is a catalog entry for the output, then it is simply saved to the location specified in the catalog. Any particular reason why you need access to the data in the hook? I could suggest workarounds accordingly.
For example, you could try omitting the catalog entry for the dataset to access it in the after_pipeline_run hook or return a copy in the node as a MemoryDataset so it gets saved + is available in the hook.