FR: More useful parsing of output by parse_check()

[MichaelChirico]

Currently, parse_check() only provides some rudimentary parsing of the 00check.log, e.g. a NOTE just becomes the full string found in the log, e.g.

writeLines(check$notes)

output:

checking if this is a source package ... NOTE
Found the following apparent object files/libraries:
  inst/libs/pkg.so
Object files/libraries should not be included in a source package.

checking dependencies in R code ... NOTE
Packages in Depends field not imported from:
  ‘int64’ ‘rcmdcheck’
  These packages need to be imported from (in the NAMESPACE file)
  for when this namespace is loaded but not attached.
Unexported object imported by a ':::' call: ‘data.table:::.shallow’
  See the note in ?`:::` about the use of this operator.

checking R code for possible problems ... NOTE
WritePkg.list: no visible binding for global variable ‘is’
Undefined global functions or variables:
  is
Consider adding
  importFrom("methods", "is")
to your NAMESPACE file (and ensure that your DESCRIPTION Imports field contains 'methods').

I think this would be more useful if we did better text parsing of the NOTEs, e.g. this could be a data.frame (or tibble/etc) with two columns: (1) source of note (here, 'checking if this is a source package', 'checking dependencies in R code', 'checking R code for possible problems', something like that), (2) a list column with content describing the note (here, [[1]] 'apparent object file', 'inst/libs/pkg.so' [[2]][[1]] 'Depends not imported', 'int64', 'rcmdcheck' [[2]][[2]] '::: call', 'data.table:::.shallow' [[3]] 'Undefined global function', 'is')

On the downside: (1) tools/R/check.R is huge, so the potential amount of parsing code to be written is also huge; and (2) it might be fragile to write text parsing as the check.R code changes over time.

For that, I'd say -- only writing parsers for the most common cases would still be great.

[gaborcsardi]

I agree that this would be great, but yeah, this does seem like a lot of maintenance work. So at this point I would say that it is unlikely that we will do this.

[MichaelChirico]

FWIW we found tools::check_packages_in_dir_details() at least does the step of peeling off the "check type" as the Check column. that part at least would be easy... it's parsing the Output column that's the real bear

[gaborcsardi]

Maybe, but we would need to double check even tools::check_packages_in_dir_details() because it only works for the current R version for sure, whereas our parser should be able to parse output from other versions.

[MichaelChirico]

It looks like that Check column is very simple text processing...

checking if this is a source package ... NOTE --> Check = "if this is a source package"
checking dependencies in R code ... NOTE --> Check = "dependencies in R code"
checking R code for possible problems ... NOTE --> Check = "R code for possible problems"

Which should not depend much on the version