Function that removes any trash in PDF after last %%EOF marker

pdf_fixer.py

import io
import os


EOF_REVERSED = b'%%EOF'[::-1]


def fix_PDF_EOF(content: io.BytesIO) -> io.BytesIO:
    """ Truncate data from last %%EOF in given PDF """

    if not content.seekable():
        raise ValueError('Content must support random access')

    initial_position = content.tell()

    # move stream position to the end
    content.seek(0, os.SEEK_END)

    # initialize values
    buffer = b''
    next_byte = EOF_REVERSED[len(buffer): len(buffer) + 1]
    position = resize_position = content.tell()

    while position > 0:

        position -= 1
        content.seek(position)
        current_byte = content.read(1)

        if next_byte == current_byte:
            # append byte to buffer
            buffer += current_byte
            if len(buffer) == len(EOF_REVERSED):
                resize_position = position
                break
        else:
            buffer = b''

        next_byte = EOF_REVERSED[len(buffer): len(buffer) + 1]

    # truncate data after %%EOF
    content.truncate(resize_position + len(EOF_REVERSED))

    # move stream to initial position
    content.seek(initial_position)
    return content


# tests.py =====================================================
import io
import pytest


@pytest.mark.parametrize('content, expected', [
    (b'%%EOF', b'%%EOF'),
    (b'%%EOF   ', b'%%EOF'),
    (b'   %%EOF ', b'   %%EOF'),
    (b'  %%EOF', b'  %%EOF'),
    (b'', b''),
    (b'%%EOFB', b'%%EOF'),
    (b'%%EOF%%%EOFb', b'%%EOF%%%EOF'),
    (b'%%EO', b'%%EO'),
    (b' %%EOF', b' %%EOF'),
    (b'%B%EOF', b'%B%EOF'),
])
def test_normal(content, expected):
    assert fix_PDF_EOF(content=io.BytesIO(content)).read() == expected