Select branch according to input file extension

KNIME Analytics Platform
1

Hello,

I would like to find a way to push data across different branches of a workflow, depending on the type of the data.

I have a File Upload node which limits the type of files to SD or to ones containing SMILES strings by setting the Valid File Extensions to "sdf, smi".

How could I send data to either an SDF Reader or Line Reader node, depending on the extension?

I have tried a CASE Switch node and managed to select the output according to the file extension. However, I have not found a node that could feed it the "raw" file content without interpreting it. Ideally, I would like to pass the content of the file to it, without having it split by line or by SD block first.

I have also tried feeding the data directly to both an SDF Reader and a Line Reader node, trying to join the output via a Joiner node with an outer join. This will not work because the Line Reader does return a table, even when SD file content is fed to it (expected behaviour, I am not complaining).

Instead of a Joiner node, I have tried an End If node (also tried with an End CASE) but I get an error: "The structures of both active input data tables are not compatible".

There must be a simple way I have not thought of yet... Anyone wishing to point me in the right direction?

Cheers,

Fred

2

Hi Fred, great example! It's tricky but possible. The problem is that the Reader nodes need to be part of the CASE Start - End construct and need to be controlled by a variable that is switched on/off depending on the file extension. This CASE Start node need to have some "fake" data as well as a variable to switch on/off the different out-ports. This variable is generated by the Java Edit Variable node. Please see the attached example that does exactly, hopefully, what you want.

3

Hello Thomas,

Thanks a lot for taking the time to provide me with an example. This made me realise that the trick was to feed the CASE a fake table and, of course, as you mentioned, to integrate the readers within the CASE structure.

I am sending you my implementation of an SD/SMILES file reader so that you can post it here if you think it may help others.

Thanks again,

Fred

4

Thanks Fred for providing the example switching SD/SMILES reading depending on the file extension. I am uploading your workflow with this post.

5

Hello,

One minor edit: I have added a Column Rename in the SMILES branch of the meta node, so that the output table always has a 'Molecule' column.

Thank you for your help,

Fred