#!/usr/bin/env python

import sys
import argparse

from hocr.parse import hocr_page_iterator
from hocr.text import hocr_page_text


def process_file(hocrfile):
    page_count = 0

    for page in hocr_page_iterator(hocrfile):
        text = hocr_page_text(page)
        sys.stdout.write(text)

        page_count += 1


if __name__ == '__main__':
    parser = argparse.ArgumentParser(description='hOCR to plaintext')
    parser.add_argument('-f', '--infile', help='Filename to read',
                        type=str, default=None)
    args = parser.parse_args()

    process_file(args.infile)
