I’ve been familiarising myself with WebMethods over the last few days. One task I have is to convert a flat file schema defined in BizTalk to one in webMethods.
At first blush I think what I want to achieve is not possible using the FF Schema tools. I have a record (with no record identifier) which contains all data on a single line. It is column delimited.
An example would be something like this:
SMITH JOHN 20080504YYNGREENRED BLUE ORNGE DOE JANE 20061221YNYRED PRPLEGREY PINK KEITTELHARVEY20051221NNYINDIGYELLOLIME PUCE PINK RED WHITE
So the first 6 fields, (surname, firstname, date, flag1, flag2, flag3) are essentially header. Everything that follows is an ‘array’ of colours. I suppose you could call it detail, which clearly repeats.
The way this structure has been done in biztalk is to define the colours column as repeating group for some arbitrary number of occurrences (although I think it can be unbounded as well). Thus it only has to be defined once, all the colours are picked up and can be treated as a stringList (to use wm terminology).
I can’t see a way to do that using the FFSchema. Never mind the idea that repeating groups in a flat file structure is a little incongruous in the first place.
The way I think I am going to do it is to define a record in a FFDict which extract the header, then a long string field to capture the remaining data for the row.
Then I will take the node that represents the remaining data and map it against another FFSchema.
Then join it all together and feed it into a canonical form.
I’m probably mixing terminology here but does this approach make sense? Or is there some aspect about ff schemas that I have missed?