1 from lxml import etree

     2 from optparse import OptionParser #@UnresolvedImport

     3 

     4 def get_options():

     5 

     6     parser = OptionParser()

     7 

     8     parser.add_option("-f", "--file", dest="outputfile",

     9                       help="destination filename", metavar="FILE", default="twitter_export_conf.xml")

    10     parser.add_option("-i", "--input", dest="inputfile", 

    11                       help="inputfile", metavar="INPUT", default=None)

    12 

    13     return parser.parse_args()

    14 

    15 if __name__ == "__main__":

    16     (options, args) = get_options()

    17     

    18     dest_filename = options.outputfile

    19     

    20     path_list = []

    21     if options.inputfile is None:

    22         path_list = args

    23     else:

    24         with open(options.inputfile, 'r') as fi:

    25             path_list = fi

    26     

    27     

    28     root = etree.Element("twitter_export")

    29     

    30     

    31     for path in path_list:

    32         

    33         iri_doc = etree.parse(path)

    34         media_nodes = iri_doc.xpath("/iri/body/medias/media[@id='video']/video")

    35         duration = int(media_nodes[0].get("dur"))/1000

    36         

    37         file_elem = etree.SubElement(root, "file")

    38         etree.SubElement(file_elem, "path").text = path

    39         etree.SubElement(file_elem, "start_date")

    40         etree.SubElement(file_elem, "duration").text = unicode(duration)

    41          

    42     tree = etree.ElementTree(root)

    43     tree.write(dest_filename, encoding="utf-8", pretty_print=True, xml_declaration=True)