Mark, I understand you are just experimenting, and that is fine.
But I have to say that using OCR for your use case is going to be the slowest and least reliable method.
Of all of the methods I've seen presented (and I haven't studied any of them in detail), I suggest trying @tom's UI scripting. I'm a bit concerned about being in a loop of 8,000 records, but if you test with 10 records or so, and put a good pause (say 0.2 sec) between each record, then I think you should be OK,
But as the other's mentioned, be sure to make a backup of your BookPedia just before you start testing and before you start a production run.
Some other thoughts:
- As part of the UI scripting, do a save every few records.
- Make sure the script does a test to ensure that the data needs to be converted; else just skip to the next record (but issue a log statement for the skipped record).
- The more I think about this, the more I'd suggest running the script from Script Debugger 7, rather than KM. This will give you more control, and make it easier to make changes to the script and test.
Let us know if you have any questions.