There is no Sinhalese traineddata file, no one has published trained data for tesseract yet. There is a 
sinhala ocr developed by UCSC, but their traineddata file is not accessible. You can find Sinhalese traineddate file from this 
sinhala ocr but it is lack of accuracy. I am looking forward to train tesseract for Sinhalese (especially for the letters in old newspapers which don't have exact fonts). I'll post here if I succeed with training Sinhalese. Anyone has knowledge about training tesseract for Sinhalese in high accuracy please comment here or share training files.