According to RFC 4180 "Common Format and MIME Type for Comma-Separated Values (CSV) Files" the quotation-character and the escape-character are both double quotes ("). In this case CVSParser does not read lines correctly:
Example cvs content:
In this case opencvs reads two datasets (first: abc, de - second: f, ghi).
Problem seems to be that CVSParser.parseLine first checks if a read character is an escape character and then, if it is a quotation character. If one interchanges those two conditions (like in the attached file), it seems to work fine.