def xmlvalidate(log):
import libxml2
from StringIO import StringIO
from random import random
prefix="...%s..." % str(random()).replace('0.','')
msg=[]
libxml2.registerErrorHandler(lambda msg,str: msg.append(str), msg)
input = libxml2.inputBuffer(StringIO(xmlEncoding.asUTF8(aString)))
reader = input.newTextReader(prefix)
reader.SetParserProp(libxml2.PARSER_VALIDATE, 1)
ret = reader.Read()
while ret == 1: ret = reader.Read()
msg=''.join(msg)
for line in msg.splitlines():
if line.startswith(prefix): log(line.split(':',4)[-1].strip())
def _validate(aString, firstOccurrenceOnly, loggedEvents, base, encoding, selfURIs=None):
"""validate RSS from string, returns validator object"""
from xml.sax import make_parser, handler
from base import SAXDispatcher
from exceptions import UnicodeError
from cStringIO import StringIO
# By now, aString should be Unicode
source = InputSource()
source.setByteStream(StringIO(xmlEncoding.asUTF8(aString)))
validator = SAXDispatcher(base, selfURIs or [base], encoding)
validator.setFirstOccurrenceOnly(firstOccurrenceOnly)
validator.loggedEvents += loggedEvents
# experimental RSS-Profile draft 1.06 support
validator.setLiterals(re.findall('&(\w+);',aString))
xmlver = re.match("^<\?\s*xml\s+version\s*=\s*['\"]([-a-zA-Z0-9_.:]*)['\"]",aString)
if xmlver and xmlver.group(1)<>'1.0':
validator.log(logging.BadXmlVersion({"version":xmlver.group(1)}))
try:
from xml.sax.expatreader import ExpatParser
class fake_dtd_parser(ExpatParser):
def reset(self):
ExpatParser.reset(self)
self._parser.UseForeignDTD(1)
parser = fake_dtd_parser()
except:
parser = make_parser()
parser.setFeature(handler.feature_namespaces, 1)
parser.setContentHandler(validator)
parser.setErrorHandler(validator)
parser.setEntityResolver(validator)
if hasattr(parser, '_ns_stack'):
# work around bug in built-in SAX parser (doesn't recognize xml: namespace)
# PyXML doesn't have this problem, and it doesn't have _ns_stack either
parser._ns_stack.append({'http://www.w3.org/XML/1998/namespace':'xml'})
def xmlvalidate(log):
import libxml2
from StringIO import StringIO
from random import random
prefix="...%s..." % str(random()).replace('0.','')
msg=[]
libxml2.registerErrorHandler(lambda msg,str: msg.append(str), msg)
input = libxml2.inputBuffer(StringIO(xmlEncoding.asUTF8(aString)))
reader = input.newTextReader(prefix)
reader.SetParserProp(libxml2.PARSER_VALIDATE, 1)
ret = reader.Read()
while ret == 1: ret = reader.Read()
msg=''.join(msg)
for line in msg.splitlines():
if line.startswith(prefix): log(line.split(':',4)[-1].strip())
validator.xmlvalidator=xmlvalidate
try:
parser.parse(source)
except SAXException:
pass
except UnicodeError:
import sys
exctype, value = sys.exc_info()[:2]
validator.log(logging.UnicodeError({"exception":value}))
if validator.getFeedType() == TYPE_RSS1:
try:
from rdflib.syntax.parsers.RDFXMLHandler import RDFXMLHandler
class Handler(RDFXMLHandler):
ns_prefix_map = {}
prefix_ns_map = {}
def add(self, triple): pass
def __init__(self, dispatcher):
RDFXMLHandler.__init__(self, self)
self.dispatcher=dispatcher
def error(self, message):
self.dispatcher.log(InvalidRDF({"message": message}))
source.getByteStream().reset()
parser.reset()
parser.setContentHandler(Handler(parser.getContentHandler()))
parser.setErrorHandler(handler.ErrorHandler())
parser.parse(source)
except:
pass
return validator
