fixhenlepdf.py

"""Remove margins from PDF generated by Henle app."""

import argparse
from pathlib import Path

from pikepdf import (Array, Pdf, PdfMatrix, parse_content_stream,
                     unparse_content_stream)


__version__ = '1.0.0'


def main():
    parser = argparse.ArgumentParser("fixhenldpdf",
                                     description="Remove margins from Henle notes PDF files")
    parser.add_argument("file", type=Path,
                        help="The original PDF file to fix")
    parser.add_argument("--output", "-o", type=Path,
                        help="Result file where to save printable PDF")
    parser.add_argument("--scale", type=float, default=1.19,
                        help="Scale (default %(default)s)")
    parser.add_argument("--xoff", type=float, default=10,
                        help="Offset X by (default %(default)s)")
    parser.add_argument("--yoff", type=float, default=-50,
                        help="Offset Y by (default %(default)s)")


    args = parser.parse_args()

    if args.output is None:
        args.output = args.file.with_stem(args.file.stem + '_print')

    with Pdf.open(args.file) as pdf:
        for page in pdf.pages:
            x = page.MediaBox[2]
            y = page.MediaBox[3]
            content = parse_content_stream(page)
            original = PdfMatrix(content[4][0])
            new = original.translated(-x*(args.scale-1)/2+args.xoff, -y*(args.scale-1)/2+args.yoff).scaled(args.scale, args.scale)
            content[4] = Array([*new.shorthand]), content[4][1]
            page.Contents = pdf.make_stream(unparse_content_stream(content))

        pdf.save(args.output)


if __name__ == '__main__':
    main()