The IBM print file format is AFP.
Advanced Function Presentation (AFP) is a presentation architecture and family of associated printer software and hardware that provides for document and information presentation independent of specific applications and devices.
This is a binary, record based format, for high volume printing.
There are tools that can manipulate these print streams. Which have you had experience with? What is it good/bad at?
Hand crafting the file in your favourite languages binary utilities.
We use StreamServe for generating AFP, but I know that they also have a component called AFPIN, that allows importing and processing existing AFP streams. I worked so far with the PDFIN component, which does the same for PDF files. Here's an excerpt from the online help for the AFPIN component:
The AFPIN filter converts AFP input to Layout eXchange Format (LXF), and enables Streamserver to identify and extract AFP formatted input documents.
LXF is a XML based intermediate format that can be transformed by StreamServe to a lot of other formats, like PDF, TIFF, PCL, PS, AFP, etc. You also can hide or replace content, e.g. overlays, OMRs and barcodes. Another option would be to store AFP document in the post-processor repository (PPR) using AFP document metadata. Coming from the PPR, the AFP documents could be sorted, split into multiple files or combined with output from other sources.
Solimar systems has tools. SPDE is one. Though I think it is very expensive. It acts as a printer so you can go from multiple printstreams to pdf, postscript, etc.