How accurate should tesseract be for a very restricted language?

[Matt Welborn]

Hi all,

I'm trying to OCR deck lists from the game Hearthstone. One such deck list is below:

I think this should be a good case for Tesseract because the set of possible lines of text is very restricted. That is, there are only 877 possible card names (e.g. "Blood to Ichor", "Slam") that can appear on a given line. However, before trying to figure out how to train Tesseract for these 877 cards and fairly unusual font, I wanted to ask: is this a good idea? Should Tesseract perform very accurately for such a task?

Thanks!

Matt