[Koha] bulkmarcimport and unicode data

hi All, I am trying to import MARC data (in unicode Tamil) into koha using bulkmarcimport.pl However, bulkmarcimport.pl does not like 3 characters (U+0B88, U+0B89, U+0BC8). Whenever these 3 characters are encountered, the rest of the string (subfield value) is gobbled and is missing from the database. But, bulkmarcimport.pl imports correctly if these 3 characters do not occur in the MARC data to be imported. Any idea whats going on ? Thanks, hari.