Turn invalid XML output from hpacucli into valid XML

hp_diagnostic_xml.py

#!/usr/bin/env python

from xml.dom.minidom import parse, parseString

import os, sys, argparse, subprocess, uuid, re, time, getpass

def hp_diagnostic_xml(output_filename):
    try:
        # Run a diagnostic, outputting to a temporary file

        while True:
            temp_filename = "/tmp/tmp%x.xml" % (uuid.uuid4())

            if not os.path.exists(temp_filename):
                break

        proc = subprocess.Popen(
            'hpacucli ctrl all diag file=%s xml=on' % (temp_filename),
            shell=True, stdout=subprocess.PIPE, stderr=subprocess.STDOUT)

        (proc_out, proc_err) = proc.communicate()

        if proc.returncode != 0:
            sys.exit("hpacucli failed with error %d: %s", proc.returncode,
                     proc_out)

        fp = open(temp_filename, 'r')
        raw_xml = fp.read()
        fp.close()

        # hpacucli doesn't escape ampersands in values sometimes, so we have to
        # make sure it does that before passing its output to the parser

        value_ampersand_regex = re.compile('value="(.*?)&(?!amp;)(.*?)"')

        def escape_matched_value(match_obj):
            return 'value="%s&%s"' % (
                match_obj.group(1), match_obj.group(2))

        (escaped_xml, num_escapes) = re.subn(
            value_ampersand_regex, escape_matched_value, raw_xml)

        print >>sys.stderr, "%d ampersands escaped" % (num_escapes)

        xml_doc = parseString(escaped_xml)

        # For some reason toprettyxml generates documents with a ton of
        # whitespace; strip that out before outputting
        doc_lines = [line for line in xml_doc.toprettyxml().split('\n')
                     if len(line.strip()) > 0]

        if output_filename is not None:
            with open(output_filename, 'w') as out_fp:
                for line in doc_lines:
                    out_fp.write(line)
        else:
            for line in doc_lines:
                print line

    finally:
        print >>sys.stderr, "Removing '%s'" % (temp_filename)

        if os.path.exists(temp_filename):
            os.remove(temp_filename)
            pass

def main():
    parser = argparse.ArgumentParser(
        description="Generates an XML file containing a valid version of "
        "hpacucli's XML diagnostic dump")
    parser.add_argument(
        "--output_filename", "-o",
        help=("the filename to which to write the diagnostic XML (default: "
              "writes to stdout)"),
        default=None)

    if getpass.getuser() != "root":
        sys.exit("Must run this script as root")

    args = parser.parse_args()

    return hp_diagnostic_xml(**vars(args))

if __name__ == "__main__":
    sys.exit(main())