xstc.py

xml开源解析代码.版本为libxml2-2.6.29,可支持GB3212.网络消息发送XML时很有用.

#!/usr/bin/env python

#
# This is the MS subset of the W3C test suite for XML Schemas.
# This file is generated from the MS W3c test suite description file.
#

import sys, os
import exceptions, optparse
import libxml2

opa = optparse.OptionParser()

opa.add_option("-b", "--base", action="store", type="string", dest="baseDir",
			   default="",
			   help="""The base directory; i.e. the parent folder of the
			   "nisttest", "suntest" and "msxsdtest" directories.""")

opa.add_option("-o", "--out", action="store", type="string", dest="logFile",
			   default="test.log",
			   help="The filepath of the log file to be created")

opa.add_option("--log", action="store_true", dest="enableLog",
			   default=False,
			   help="Create the log file")

opa.add_option("--no-test-out", action="store_true", dest="disableTestStdOut",
			   default=False,
			   help="Don't output test results")

opa.add_option("-s", "--silent", action="store_true", dest="silent", default=False,
			   help="Disables display of all tests")

opa.add_option("-v", "--verbose", action="store_true", dest="verbose",
			   default=False,
			   help="Displays all tests (only if --silent is not set)")

opa.add_option("-x", "--max", type="int", dest="maxTestCount",
			   default="-1",
			   help="The maximum number of tests to be run")

opa.add_option("-t", "--test", type="string", dest="singleTest",
			   default=None,
			   help="Runs the specified test only")
			   
opa.add_option("--tsw", "--test-starts-with", type="string", dest="testStartsWith",
			   default=None,
			   help="Runs the specified test(s), starting with the given string")

opa.add_option("--rieo", "--report-internal-errors-only", action="store_true",
			   dest="reportInternalErrOnly", default=False,
			   help="Display erroneous tests of type 'internal' only")

opa.add_option("--rueo", "--report-unimplemented-errors-only", action="store_true",
			   dest="reportUnimplErrOnly", default=False,
			   help="Display erroneous tests of type 'unimplemented' only")

opa.add_option("--rmleo", "--report-mem-leak-errors-only", action="store_true",
			   dest="reportMemLeakErrOnly", default=False,
			   help="Display erroneous tests of type 'memory leak' only")

opa.add_option("-c", "--combines", type="string", dest="combines",
			   default=None,
			   help="Combines to be run (all if omitted)")
			   
opa.add_option("--csw", "--csw", type="string", dest="combineStartsWith",
			   default=None,
			   help="Combines to be run (all if omitted)")			   

opa.add_option("--rc", "--report-combines", action="store_true",
			   dest="reportCombines", default=False,
			   help="Display combine reports")

opa.add_option("--rec", "--report-err-combines", action="store_true",
			   dest="reportErrCombines", default=False,
			   help="Display erroneous combine reports only")

opa.add_option("--debug", action="store_true",
			   dest="debugEnabled", default=False,
			   help="Displays debug messages")

opa.add_option("--info", action="store_true",
			   dest="info", default=False,
			   help="Displays info on the suite only. Does not run any test.")
opa.add_option("--sax", action="store_true",
			   dest="validationSAX", default=False,
			   help="Use SAX2-driven validation.")
opa.add_option("--tn", action="store_true",
			   dest="displayTestName", default=False,
			   help="Display the test name in every case.")

(options, args) = opa.parse_args()

if options.combines is not None:
	options.combines = options.combines.split()

################################################
# The vars below are not intended to be changed.
#

msgSchemaNotValidButShould =  "The schema should be valid."
msgSchemaValidButShouldNot = "The schema should be invalid."
msgInstanceNotValidButShould = "The instance should be valid."
msgInstanceValidButShouldNot = "The instance should be invalid."
vendorNIST = "NIST"
vendorNIST_2 = "NIST-2"
vendorSUN  = "SUN"
vendorMS   = "MS"

###################
# Helper functions.
#
vendor = None

def handleError(test, msg):
	global options
	if not options.silent:
		test.addLibLog("'%s'   LIB: %s" % (test.name, msg))
	if msg.find("Unimplemented") > -1:
		test.failUnimplemented()
	elif msg.find("Internal") > -1:
		test.failInternal()
		
	
def fixFileNames(fileName):
	if (fileName is None) or (fileName == ""):
		return ""
	dirs = fileName.split("/")
	if dirs[1] != "Tests":
		fileName = os.path.join(".", "Tests")
		for dir in dirs[1:]:
			fileName = os.path.join(fileName, dir)	
	return fileName

class XSTCTestGroup:
	def __init__(self, name, schemaFileName, descr):
		global vendor, vendorNIST_2
		self.name = name
		self.descr = descr
		self.mainSchema = True
		self.schemaFileName = fixFileNames(schemaFileName)
		self.schemaParsed = False
		self.schemaTried = False

	def setSchema(self, schemaFileName, parsed):
		if not self.mainSchema:			
			return
		self.mainSchema = False
		self.schemaParsed = parsed
		self.schemaTried = True

class XSTCTestCase:

		   # <!-- groupName, Name, Accepted, File, Val, Descr
	def __init__(self, isSchema, groupName, name, accepted, file, val, descr):
		global options
		#
		# Constructor.
		#
		self.testRunner = None
		self.isSchema = isSchema
		self.groupName = groupName
		self.name = name
		self.accepted = accepted		
		self.fileName = fixFileNames(file)
		self.val = val
		self.descr = descr
		self.failed = False
		self.combineName = None

		self.log = []
		self.libLog = []
		self.initialMemUsed = 0
		self.memLeak = 0
		self.excepted = False
		self.bad = False
		self.unimplemented = False
		self.internalErr = False
		self.noSchemaErr = False
		self.failed = False
		#
		# Init the log.
		#
		if not options.silent:
			if self.descr is not None:
				self.log.append("'%s'   descr: %s\n" % (self.name, self.descr))		
			self.log.append("'%s'   exp validity: %d\n" % (self.name, self.val))

	def initTest(self, runner):
		global vendorNIST, vendorSUN, vendorMS, vendorNIST_2, options, vendor
		#
		# Get the test-group.
		#
		self.runner = runner
		self.group = runner.getGroup(self.groupName)				
		if vendor == vendorMS or vendor == vendorSUN:
			#
			# Use the last given directory for the combine name.
			#
			dirs = self.fileName.split("/")
			self.combineName = dirs[len(dirs) -2]					
		elif vendor == vendorNIST:
			#
			# NIST files are named in the following form:
			# "NISTSchema-short-pattern-1.xsd"
			#						
			tokens = self.name.split("-")
			self.combineName = tokens[1]
		elif vendor == vendorNIST_2:
			#
			# Group-names have the form: "atomic-normalizedString-length-1"
			#
			tokens = self.groupName.split("-")
			self.combineName = "%s-%s" % (tokens[0], tokens[1])
		else:
			self.combineName = "unkown"
			raise Exception("Could not compute the combine name of a test.")
		if (not options.silent) and (self.group.descr is not None):
			self.log.append("'%s'   group-descr: %s\n" % (self.name, self.group.descr))
		

	def addLibLog(self, msg):		
		"""This one is intended to be used by the error handler
		function"""
		global options		
		if not options.silent:
			self.libLog.append(msg)

	def fail(self, msg):
		global options
		self.failed = True
		if not options.silent:
			self.log.append("'%s' ( FAILED: %s\n" % (self.name, msg))

	def failNoSchema(self):
		global options
		self.failed = True
		self.noSchemaErr = True
		if not options.silent:
			self.log.append("'%s' X NO-SCHEMA\n" % (self.name))

	def failInternal(self):
		global options
		self.failed = True
		self.internalErr = True
		if not options.silent:
			self.log.append("'%s' * INTERNAL\n" % self.name)

	def failUnimplemented(self):
		global options
		self.failed = True
		self.unimplemented = True
		if not options.silent:
			self.log.append("'%s' ? UNIMPLEMENTED\n" % self.name)

	def failCritical(self, msg):
		global options
		self.failed = True
		self.bad = True
		if not options.silent:
			self.log.append("'%s' ! BAD: %s\n" % (self.name, msg))

	def failExcept(self, e):
		global options
		self.failed = True
		self.excepted = True
		if not options.silent:
			self.log.append("'%s' # EXCEPTION: %s\n" % (self.name, e.__str__()))

	def setUp(self):
		#
		# Set up Libxml2.
		#
		self.initialMemUsed = libxml2.debugMemory(1)
		libxml2.initParser()
		libxml2.lineNumbersDefault(1)
		libxml2.registerErrorHandler(handleError, self)

	def tearDown(self):
		libxml2.schemaCleanupTypes()
		libxml2.cleanupParser()
		self.memLeak = libxml2.debugMemory(1) - self.initialMemUsed

	def isIOError(self, file, docType):
		err = None
		try:
			err = libxml2.lastError()
		except:
			# Suppress exceptions.
			pass
		if (err is None):
			return False
		if err.domain() == libxml2.XML_FROM_IO:
			self.failCritical("failed to access the %s resource '%s'\n" % (docType, file))

	def debugMsg(self, msg):
		global options
		if options.debugEnabled:
			sys.stdout.write("'%s'   DEBUG: %s\n" % (self.name, msg))

	def finalize(self):
		global options
		"""Adds additional info to the log."""
		#
		# Add libxml2 messages.
		#
		if not options.silent:
			self.log.extend(self.libLog)
			#
			# Add memory leaks.
			#
			if self.memLeak != 0:
				self.log.append("%s + memory leak: %d bytes\n" % (self.name, self.memLeak))

	def run(self):
		"""Runs a test."""
		global options

		##filePath = os.path.join(options.baseDir, self.fileName)
		# filePath = "%s/%s/%s/%s" % (options.baseDir, self.test_Folder, self.schema_Folder, self.schema_File)
		if options.displayTestName:
			sys.stdout.write("'%s'\n" % self.name)
		try:
			self.validate()
		except (Exception, libxml2.parserError, libxml2.treeError), e:
			self.failExcept(e)
			
def parseSchema(fileName):
	schema = None
	ctxt = libxml2.schemaNewParserCtxt(fileName)
	try:
		try:
			schema = ctxt.schemaParse()
		except:
			pass
	finally:		
		del ctxt
		return schema
				

class XSTCSchemaTest(XSTCTestCase):

	def __init__(self, groupName, name, accepted, file, val, descr):
		XSTCTestCase.__init__(self, 1, groupName, name, accepted, file, val, descr)

	def validate(self):
		global msgSchemaNotValidButShould, msgSchemaValidButShouldNot