Parsing a mailaddress as in https://issues.apache.org/jira/browse/MIME4J-31 results in a ParseException. Parsing a mailaddress starting with a dot (.) results in a ParseException.
When parsing an addressfield with multiple adresses, the Exception occuring while parsing a single address is caught and null is returned as the resulting addresslist. (this breaks tika as it expects an empty list rather than null)
It would be nice if invalid addresses would be handled more gracefully when in lenient mode. And it would be nice if at least the correct addresses would be returned while parsing an addresslist with a corrupted address.
I am using Mime4J via the Apache Tika project to extract text from emails for indexing in Lucene. The textstream of tika is directly read by a lucene field and indexing fails if an exception is thrown by Mime4J. This currently happens every time a headerfield contains more than 1000 characters due to tika using the unusable mime4j standardconfiguration ( https://issues.apache.org/jira/browse/TIKA-640 ), and every time a malformed emailaddress is encountered ( https://issues.apache.org/jira/browse/TIKA-641 ).
These problems can be taken care of in Tika, but there is no way for Tika to retrieve the working mailaddresses out of a list, if Mime4j returns only none; maybe this problem could be addressed in Mime4J.