opencdms-process: create pytest tests that verify calls to R functions generate the expected R warnings

[lloyddewit]

• Python version: 3.8.10
• R version: 4.1.2
• Operating System: Ubuntu 20.04

I want to write a 'PyTest' function that checks if a specific call to an R function raises the expected warning(s). For example, the call below should generate warning 'facet_by set to none since no stations are given in data':

    cdms_products.histogram_plot(
        path=output_path_actual,
        file_name=file_name_actual,
        data=niger50,
        date_time="date",
        elements=["tmax"],
        facet_by="stations",
    )

I am using the 'rpy2' library to call the R function. The 'rpy2' library does not yet seem to allow warnings/errors generated in the R environment to be read in the Python environment. I created 'rpy2' issue 894. This describes the problem in detail.

I see two possible ways ahead:

1. Wait to see if the rpy2 issue is resolved.
2. Use a workaround like writing the R output to file in the R environment, (example here).

I prefer option 1.

Personal notes on possible approaches if/when 'rpy2' issue above is resolved

The Python wrapper would write any R warnings/errors to a stream that is returned to the calling function. Or it could write errors/warnings to a log file. The log file name would be the same as the output file name, except that the file name extension would be ".log".
Example code (Python wrapper function):

    from rpy2.rinterface_lib.callbacks import logger as rpy2_logger
    file_name_log = os.path.join(path, os.path.splitext(file_name)[0] + ".log")
    log_file_handler = logging.FileHandler(file_name_log)
    rpy2_logger.addHandler(log_file_handler)

Example code ('pytest' function):

    cdms_products.histogram_plot(
        path=output_path_actual,
        file_name=file_name_actual,
        data=niger50,
        date_time="date",
        elements=["tmax"],
        facet_by="stations",
    )
    assert __is_expected_file(file_name_actual)
    assert __is_expected_file(os.path.splitext(file_name_actual)[0]+".log")

Alternatively, the Python wrapper function could ignore the R warnings, and the 'pytest' function could analyse the warnings directly (see pytest warning documentation here and here).