cnxmlplus-validate

#!/usr/bin/env python
from lxml import etree
import sys, os
import argparse
import inspect

from XmlValidator import XmlValidator

SPEC_PATH = os.path.join(os.path.dirname(inspect.getfile(XmlValidator)))
MY_PATH = os.path.realpath(os.path.dirname(__file__))

# Parse command line arguments
argumentParser = argparse.ArgumentParser(description='Check the validity of a CNXML+ document.')
argumentParser.add_argument(
    '--spec', dest='specFilename',
    default='spec.xml',
    help='Filename of the XML specification document.')
argumentParser.add_argument(
    '--clean-up', dest='produceCleanedXML', action='store_true',
    help='Write to standard output a cleaned-up version of the input XML. Will delete all offending elements.')
argumentParser.add_argument(
    '-o', dest='outputFilename',
    help='Write output to given filename rather than stdout.')
argumentParser.add_argument(
    'filename', nargs='+',
    help='One or more filenames to process.')
commandlineArguments = argumentParser.parse_args()

isSingleFile = (len(commandlineArguments.filename) == 1)

if commandlineArguments.outputFilename is None:
    outputFile = sys.stdout
else:
    outputFile = open(commandlineArguments.outputFilename, 'wt')
xmlValidator = XmlValidator(open(os.path.join(SPEC_PATH, commandlineArguments.specFilename),'rt').read())

cleanUp = commandlineArguments.produceCleanedXML
for filename in commandlineArguments.filename:
    if not isSingleFile:
        sys.stderr.write(filename + '\n')

    xmlValidator.validate(
        open(filename,'rt').read(),
        iCleanUp=cleanUp)

    if cleanUp:
        print etree.tostring(xmlValidator.dom, xml_declaration=True, encoding='utf-8')