1. Are the output records to be a fixed length (ie 94 characters)?
2. Are the line terminations included in that length?
3. Is there _anything_ funny about the encoding? You aren't trying to do ASCII in one side and EBCDIC out the other?

Assuming the answers are 'yes', 'yes', and 'no', then the logic runs like this pseudo code:

open input_file
open output_file

while (get-next-record) {
    rec_length = length(record)

    if (rec_length > 92) {
        Throw A Fit And Exit With An Error
    } else {
        record = " " x (92 - rec_length) . record ."crlf"
        write record to output_file
    }
}
[download]

Read each record in your input, pad it on the left to 92 characters, append your line termination, and write the resultant string to a new file. Repeat until done-enough.

update (spelling too): or just use substr