Ability to delete a group

ocrd workspace delete-group could be a command to delete a group/fileGrp in the workspace/METS file.

My use case for this:

Currently, I use xmlstarlet to remove a group of files so that I can repeat a script run, i.e. to enable repeated (and almost idempotent) runs of the following:

remove_filegrp() {
  filegrp_use=$1
  mets=$2

  xmlstarlet ed --inplace \
    -N mets=http://www.loc.gov/METS/ \
    -d "//mets:fileGrp[@USE='$filegrp_use']" $mets
}

remove_filegrp OCR-D-SEG-LINE mets.xml
ocrd-ocropy-segment -l DEBUG -m mets.xml -I OCR-D-IMG -O OCR-D-SEG-LINE

[kba]

I hear you, I’m on it. This, the TEMP-directory-on-validation and input-file-copying, and improving ocrd process are next on my agenda for core.

[bertsky]

Like an option --overwrite/–force for ocrd workspace add?

We already have --force (“If file with ID already exists, replace it”). What behavior changes do you want here?

No, I meant an option for ocrd process and all individual module processor CLIs. (The --force option is not exposed to those CLIs.)

Thus we can have repeated runs without even looking at which files and groups are affected.

OK then, ocrd workspace rm-group shouldn’t be too hard to implement. Should this behave like rmdir and only delete empty fileGroups or like rm -r and also delete non-empty fileGroups?

Probably should also have the equivalent for removing files then, i.e. ocrd workspace rm. What do you think?

IMHO, this depends on the above: if we get a simple way to repeat processing steps, then these removal operations would become but an extra, for special circumstances. So it would be okay to be cautious and have the user first delete all files (and file references!) – like rm – and then empty groups – like rmdir.

But if (for some reason) we don’t get some ocrd process -f, then removal should be easy to use – like rm -fr.