Thanks guys for the feedback.
I was of the assumption that there was no reliable way to guess the structure of the data but wanted to come here where I could find people much smarter than me and get their feedback (and confirmation)
The data is coming out of an old *ty system and only allows export via HTML attached to email - That is why our process exists - to take their *ty, unusable data format and convert it to CSV so that they can manipulate and report on it.
- Field lengths can't be fixed - they can be changed at anytime (and do get changed)
Ideally, I want to develop this to be intuitive - get an end user to just build a new report - send it out to our email and get a CSV back... But I just can't see how without knowing field header lengths...
Thanks for your time.