Get OVF disk

disk.py

from six.moves.urllib import parse
from lxml import etree


SPECS = {
    'http://www.vmware.com/interfaces/specifications/vmdk.html': 'vmdk',
    'https://people.gnome.org/~markmc/qcow-image-format.html': 'qcow',
}


def get_ovf_disk(ovf):
    tree = etree.parse("file.ovf")
    root = tree.getroot()
    ovf_ns = root.nsmap['ovf']

    id_attr = '{%s}id' % ovf_ns
    href_attr = '{%s}href' % ovf_ns
    files = {f.get(id_attr): f.get(href_attr) for f in
             root.findall('ovf:References/ovf:File', root.nsmap)}

    # we do not care about more than one disk
    disk = root.find('ovf:DiskSection/ovf:Disk', root.nsmap)
    if disk is not None:
        format_attr = '{%s}format' % ovf_ns
        fileref_attr = '{%s}fileRef' % ovf_ns
        ovf_format = disk.get(format_attr)
        if not ovf_format:
            raise Exception("Expecting some format!")
        (format_url, _) = parse.urldefrag(ovf_format)
        try:
            disk_format = SPECS[format_url]
        except KeyError:
            raise Exception("Unknown format!")
        try:
            disk_file = files[disk.get(fileref_attr)]
        except KeyError:
            raise Exception("Unknown disk!")
        return (disk_format, disk_file)
    return None, None