thank u. although... i don't think it has all of the text typed by the users (eg set name, profile, etc) and it has some text that wasn't typed by users (eg ").
i think it's better to simply have a plain text dump of every memo transaction (delimited with double returns) so we can follow it and parse it correctly.