Instead of posting a csv file, use the NodeCalc (ethercalc) page to get data, parse it and create a corpus.

To test it : create a nodeCalc, copy/paste into it a CSV file corpus example and adapt the parser.