In attachment the input parser (the .py file), a simple .csv input file, and the resulting output files (I used XML so you can easily inspect them), but of course you can then use the input parser on your real input file and with Kobo output. You might want to use the sort/merge switches as well.
Code:
$ python -m penelope -i input.csv -j csv --csv-fs "\\t" -f en -t en -p xml -o output_no_input_parser.xml
$ python -m penelope -i input.csv -j csv --csv-fs "\\t" -f en -t en -p xml -o output_with_input_parser.xml --input-parser multiple_index_words.py