Self-contained archive containing solution and data to reproduce this bug can be downloaded at
http://bio.mquter.qut.edu.au/Downloads/GenBankParserBugReport.zip (5.5MB)
 
When parsing a “/note” qualifier, if there are multiple lines of note text, sometimes a line break will happen such that a "/" which is part of the text will start a continuation line. This causes the parser to treat the following text on that line as a qualifier and any following text lines as values for the qualifier. This happened in MBF v1 and still happens in v2.
An example of this is in ftp://ftp.ncbi.nih.gov/genomes/Bacteria/

        Acaryochloris_marina_MBIC11017_uid58167/NC_009925.gbk:

misc_feature 948729..949922

                /gene="kmo"
                /locus_tag="AM1_0975"
                /note="2-polyprenyl-6-methoxyphenol hydroxylase and
                related FAD-dependent oxidoreductases [Coenzyme metabolism
                / Energy production and conversion]; Region: UbiH;
                COG0654"

 
The parser identifies the following qualifiers for this feature. Observe that the /note has been truncated and a spurious qualifier has been introduced based on the tail of the /note.
 
Qualifier: { Energy production and conversion]; Region: UbiH;}
Value : { COG0654"}
 
Qualifier: {db_xref}
Value : {"CDD:30999"}
 
Qualifier: {gene}
Value : {"kmo"}
 
Qualifier: {locus_tag}
Value : {"AM1_0975"}
 
Qualifier: {note}
Value : {"2-polyprenyl-6-methoxyphenol hydroxylase and\nrelated FAD-dependent oxidoreductases [Coenzyme metabolism}