tesseract ocr training new font

Required : 1- Training tesseract 4 or 5 on Arabic and Indian numbers accurately and extract them only from non-texts in the image in a trained file . 2- Training tesseract on Arabic date and making it clear that it is a date. 3- Use opencv python to detect the orientation of the Arabic text and recycle the paper to the correct automatic format. 4- Use python Extract the logos into the document and save it in a folder.