one would expect that '''query''' produces an output file that can be read back by '''upsert'''.
this does not work, because query will duplicate the '''entities:'' key for each batch, essentially writing a corrupt yaml file.
(for yaml, but i guess other formats are similarly affected, it's probably a similiar issue for json, but i have not tried; for csv it likely does not cause issues because there is no header?)
even worse, upsert will accept such a yaml file as input, but apparently only use the last batch (because the parser internally overwrites repeating keys?)
but that is probably not a bug in dsio itself, because the yaml parser is from a library and not part of dsio.)
one would expect that '''query''' produces an output file that can be read back by '''upsert'''. this does not work, because query will duplicate the '''entities:'' key for each batch, essentially writing a corrupt yaml file. (for yaml, but i guess other formats are similarly affected, it's probably a similiar issue for json, but i have not tried; for csv it likely does not cause issues because there is no header?)
even worse,
upsert
will accept such a yaml file as input, but apparently only use the last batch (because the parser internally overwrites repeating keys?) but that is probably not a bug in dsio itself, because the yaml parser is from a library and not part of dsio.)