Unable to train IDP with unstructured document type

Hi, 

We had a requirement to upload a purchase order and get the data extracted from it. 
but the issue here is the document contains a table with a number of rows,
 
The IDP is unable to extract each row, by the way, the data is getting extracted like a bunch, 
As an example we can have a look into the screenshot,where Sno 2,3,4 are extracted as a single row.

Is there anything we can handle these kind of issues in IDP.

Thanks in Advance.