Is your feature request related to a problem? Please describe.
Currently, child-project validate corpus/ will do a series of checks that are perfect when setting up a corpus; and we have one flag that adapts the behavior of validate to the context of importing annotations by ignoring the absence of recordings.
We have other procedures that also require certain things of the corpus, which it would be great to validate to this end first:
metrics extraction procedures require annotations to be complete & local
audio processors require the raw recordings to be local
sampler requires recordings.csv and recordings of a certain profile to be local
.eaf builder requires just metadata
Describe the solution you'd like
I don't think we should create a flag for each process, but we could create flags that perform only subsections of validate. For instance child-project validate corpus/ --annotations vtc/converted/ can check that those annotations are complete & local; child-project validate corpus/ --audio raw/ converted/standard/.
I wonder whether this is not equivalent to the user doing datalad get annotations/vtc/converted/ -- I could imagine it is not, because such a command will not alert a user that a given child or recording doesn't have its corresponding vtc/converted annotation.
Is your feature request related to a problem? Please describe. Currently,
child-project validate corpus/
will do a series of checks that are perfect when setting up a corpus; and we have one flag that adapts the behavior of validate to the context of importing annotations by ignoring the absence of recordings.We have other procedures that also require certain things of the corpus, which it would be great to validate to this end first:
recordings.csv
and recordings of a certain profile to be localDescribe the solution you'd like I don't think we should create a flag for each process, but we could create flags that perform only subsections of
validate
. For instancechild-project validate corpus/ --annotations vtc/converted/
can check that those annotations are complete & local;child-project validate corpus/ --audio raw/ converted/standard/
.I wonder whether this is not equivalent to the user doing
datalad get annotations/vtc/converted/
-- I could imagine it is not, because such a command will not alert a user that a given child or recording doesn't have its correspondingvtc/converted
annotation.