[AlPe]

Hi Oren,

thanks for confirming Penelope v3.1.3 works on Windows with Python 3.x, at least now we can suggest people to use Python 3.x.

With respect to your question:

My problem is when there is two or more words to one translation.
...

I am not sure I understood it correctly. I understood that you have a CSV file (I use "<TAB>" to indicate the field separator for the example):

word1|word2<TAB>definition

where "definition" is the definition associated with both "word1" and "word2". So, you expect the output Kobo to have two entries, one with index word "word1", the other with index word "word2", both associated with "definition".

If this is the case, unfortunately Penelope cannot understand that "word1|word2" is composed by two strings, each to be associated with "definition". It sees it as a single string, exactly as if it was "word3".

However, you can write a simple input parser and use the "--input-parser" switch. (I can code it for you, if you confirm I understood it correctly) Or use a text editor and some regex to duplicate the row:

word1<TAB>definition
word2<TAB>definition

and feed Penelope with this derived file.

If I did not understand correctly, let me know.