Is it possible for Connector to communicate with and work with an IBM 5250 Emulator, has anyone had any experience with implementing this? During initial testing, it seemed as if it didn't work.
All of the emulators support internal scripting languages, and let you screen scrape without messing around with OCR. It's much faster and infinitely more accurate. Then the question is what to do with the data - we use the toolkit to send the data and search requests into Laserfiche. It's much more work overall, but does work reliably once you have made the time investment.
It doesn't even work with OCR?
It seems its working with OCR, but its proving difficult and it seems we may have found a glitch. When we test the process before publishing it is able to extract the data we want. But after we publish it and try to extract that same data it does not work. So we are messing around with regular expressions to see if we can get consistent results.
You may already have found this, but there are several image enhancements that you can do in the OCR process in connector.
We've used the image enhancements to good effect. I'm working on this case for our client and it's puzzling.
During the first steps of setting up a Connector profile, we set the source application, then use "OCR to extract text", which reads the text back (at this point) successfully. We then build a custom search using a pattern match to obtain a Customer ID value from the "ScreenText". The regular expression we are using works in the search syntax token editor, so we assume it will work when we test. However, when we run the test, there is no "Customer ID" value being obtained. We're not sure if it is the OCR process or the Regular Expression that is failing. It should return a search value of "747115", but it is returning an empty value, instead.
The RexEx is "Customer ID.*(\d\d\d\d\d\d)", but this may not be the best. I'm going to try this "\r?\n?Customer\s*ID[^\d]+(\d{6})[^\d]+" to see if it makes a difference, assuming the OCR is consistent, of course. :)