add unit tests for new diff_reports utility function

[sarabodach]

https://github.com/pharmaverse/sdtmchecks/blob/devel/R/utils.R#L602-L728

[rossfarrugia]

as per Slack message Jim Rothstein is keen to take on this issue once he gets write access.

[sarabodach]

Hi @jimrothstein - thank you for your help!

[jimrothstein]

@sarabodach Maybe not smart question, where are sdtm datasets? j from run_all_checks.R

#' # Assuming sdtm datasets are in your global environment

I do see samples AE etc in code for each test, but not AE itself.

[sarabodach]

Hi @jimrothstein - that is a good question! We have just used small test datasets in headers to illustrate the data check logic or for testing in existing unit tests. The package itself does not include larger SDTM domains in general.

I haven't tried this myself yet, but there are some pharmaverse packages (such as https://github.com/pharmaverse/pharmaversesdtm) with test SDTM data in case you are interested in seeing other examples that likely have more types of scenarios. We would not want to add this additional package as a dependency though, so I'd advise not to include this package in the actual unit test programs.

I'd suggest starting small with a few AE dataframe examples that are in check headers to see whether the list of data check results A yields expected results when comparing to the list of data check results B. The result of a check may reflect 0 records if the check function passes, so it is not necessary to have all types of data examples for creating unit tests.

[jimrothstein]

@sarabodach Greatly appreciate the replies; keeps me from going too far astray.

• Do you mind if add a few comments to code?
  I find the code clever in places (use of attributes). comments would help me, or a future person. But if most checking code is done, then I don't think worth the effort.

• paramversesdtm is pulling data from http://github.com/cdisc-org fills in missing link for me; will look over the code test.

• Create list of AE dataframe examples good idea and will do now. (I keep wanting check-*-* functions to return data.frame, not lists, but will get used to i!)

Thx again.

[sarabodach]

Hi @jimrothstein , comments are fine if they are relevant. But they can also be cleaned up later as well in case it's easier for you to move forward with them. Thanks!

[jimrothstein]

@sarabodach My goal is unit tests for diff_reports, but I notice that run_check and run_all_checks do not yet have unit tests. They build on each other. (Also forces me to go through code, which is good thing.)

We talked about constructing a list of dummy datasets. So I did that. I then used this dummy list to test run_check, for the one check_ae_aeanoth . Not yet in testthat format.

~~Could you look at code, please; I will rework code and turn it into real testhat code. Thx. (Looks like code is still on my forked repo; should be on branch 295_ . Need to figure this out. Sorry, beginner issues.)~~

Update (Monday, 17 JUNE)

• I think best if I redo branch 295_diff_reports.. and push a clean version. (Now reading github branch DOCS)