this is solutions:
we can use CasperJS, for doing bot/robot function, like navigate web, click fill form, etc, then for OCR function , i need to ask are PDF use image? or PDF text, we can use PDF2Text linux open source tool, we running this service backgroundly(serverly) , let discuss, iam so interested