Parsing/unparsing with optional characters

[Larry Barber]

Does anyone have suggestions on how structure a schema to parse a file where 
both "KEYW" and "KEYWORD" are valid and considered identical and that will 
return with the original version when unparsed?