pdf.py

@contextmanager
def convert_to_pdf(filename):
    """Creates a PDF document based on input filename.

    Returns a path to PDF file
    """
    client = docker.from_env()

    source = os.path.join('/tmp/{}'.format(os.path.basename(filename)))
    expected_filename = os.path.splitext(os.path.basename(filename))[0] + '.pdf'

    dest = os.path.join('/tmp/{}'.format(filename))
    log.debug('Source {}'.format(source))
    args = [
            '--headless',
            '--invisible',
            '--convert-to', 'pdf:writer_pdf_Export',
            '--outdir', '/tmp/',
            source
        ]
    log.info('Args {!r}'.format(args))
    container = client.containers.create(
        'libreoffice',
        args,
        tty=True)
    try:
        log.info('Created new container {}'.format(container.id))
        # Load file inside the container using its filename through a dummy tar file
        fileobj = io.BytesIO()
        with tarfile.open(fileobj=fileobj, mode='w') as tar:
            with open(filename, 'rb') as fin:
                # Create a TarInfo object based on file object
                tarinfo = tar.gettarinfo(arcname=os.path.basename(filename), fileobj=fin)
                # Set chmod to 0444 so the container user can read it from /tmp
                tarinfo.mode = stat.S_IRUSR | stat.S_IRGRP | stat.S_IROTH
                log.info('Tarinfo {}'.format(tarinfo))
                tar.addfile(tarinfo, fin)
        # Upload dummy tar into /tmp/    
        res = container.put_archive('/tmp/', fileobj.getvalue())
        # Start the container
        container.start()
        # Wait for the container to finish
        container.wait()
        # Get the logs line by line just to check how it was
        for line in container.logs().decode('utf-8').splitlines():
            log.info('Line {!r}'.format(line))
        # Download /tmp/ contents as a tar archive
        bits, archive_stat = container.get_archive('/tmp/')
        log.info('Data archive {}'.format(archive_stat))
        # This will extract the container
        with tempfile.NamedTemporaryFile(suffix='.tar') as temp_bits:
            # Download tar chunks
            for chunk in bits:
                temp_bits.write(chunk)
            temp_bits.flush()
            temp_bits.seek(0)
            # Open the downloaded tar file
            with tarfile.open(fileobj=temp_bits, mode='r') as tar:
                for member in tar.getmembers():
                    log.info('List {}'.format(member))
                member = tar.getmember(os.path.join('tmp', expected_filename))
                log.info('Member {}'.format(member))
                yield tar.extractfile(member)
    finally:
        log.info('Remove container {}'.format(container.id))
        container.remove()