Applies XSL transform to XML returned from an HTTP request, then validates it against an XSD/schema.

xml_transform_and_validate.py

from __future__ import print_function

import argparse
import lxml.etree as ET
import urllib

parser = argparse.ArgumentParser(description="Transforms and Validates XML")
parser.add_argument("--feeds", help="the text file that holds the urls to be validated", default="feeds.txt")
parser.add_argument("--log", help="the output log file that reports all errors", default="log.txt")
parser.add_argument("-v", "--verbose", help="increase output verbosity", action="store_true")
args = parser.parse_args()

def getDOM(feedUrl):
	# hit the url (fetch the xml), then parse it into a dom object
	return ET.parse(urllib.urlopen(feedUrl))

def getXslt(feedUrl):
	# examine the feedUrl, determine which XSLT to use
	if "myspecialfeeds/rest/abc" in feedUrl: return "xslts/Abc.xslt"
	if "myspecialfeeds/rest/def" in feedUrl: return "xslts/Def.xslt"
	if "myspecialfeeds/rest/ghi" in feedUrl: return "xslts/Ghi.xslt"
	return

def applyXslt(dom, xsltName):
 	# parse the xsl transform into an object, then apply it to the incoming dom
 	transform = ET.XSLT(ET.parse(xsltName))
 	return transform(dom)

def getXsd(feedUrl):
	# examine the feedUrl, determine which XSD to use
	if "myspecialfeeds/rest/abc" in feedUrl: return "xsds/Abc.xsd"
	if "myspecialfeeds/rest/def" in feedUrl: return "xsds/Def.xsd"
	if "myspecialfeeds/rest/ghi" in feedUrl: return "xsds/Ghi.xsd"
	return

def applyXsd(transformedXml, xsdName):
 	# parse the xsd into an object, then validate the incoming dom against it
 	xmlschema = ET.XMLSchema(ET.parse(xsdName))
 	xmlschema.assertValid(transformedXml)
 	return

with open(args.log, "a") as logFile:
	with open(args.feeds, "r") as feedsFile:
		print()
		for feedLine in feedsFile:
			# strip out the line terminator
			feedUrl = feedLine.rstrip("\n")
			
			# if url is commented out using '#', skip line
			if feedUrl.startswith("#"): continue
			
			# if the line is empty, skip line
			if not feedUrl: continue

			print("Validating: " + feedUrl)
			try:
				# fetch xml from feed via url
				dom = getDOM(feedUrl)

				# only print the xml if verbose logging is enabled
				if args.verbose: print(ET.tostring(dom, pretty_print=True))

				# determine which xsl transform to apply
				xslt = getXslt(feedUrl)

				# apply the xsl transform
				transformed_xml = applyXslt(dom, xslt)

				# only print the transformed xml if verbose logging is enabled
				if args.verbose: print(ET.tostring(transformed_xml, pretty_print=True))
				
				# determine which xsd to validate transformed xml against
				xsd = getXsd(feedUrl)

				# validate the xml against the xsd
				applyXsd(transformed_xml, xsd)

				print("Feed is valid.")
			except Exception, e:
				print("Feed is INVALID! (See log for details.)")
				
				print("Invalid feed URL: " + feedUrl, file=logFile)
				print(e, file=logFile)
				if args.verbose:
					print("Feed URL returned the following XML:", file=logFile)
					print(ET.tostring(dom, pretty_print=True), file=logFile)
					print("Feed URL XML transformed to:", file=logFile)
					print(ET.tostring(transformed_xml, pretty_print=True), file=logFile)
				print("", file=logFile)
				
			print()