While not exactly an answer to your question, but a help for parsing sequence formats. It seems that you are parsing EMBL files. Easier way to get sequences from those might be emboss seqret or perhaps bioperl.