def test_parse_InputSource(self):
make_xml_file(self.data, 'iso-8859-1', None)
with open(TESTFN, 'rb') as f:
input = InputSource()
input.setByteStream(f)
input.setEncoding('iso-8859-1')
self.check_parse(input)
def parse(self) :
if (isinstance(self.source,unicode)) :
# Create a string source
file = io.StringIO(self.source)
input = InputSource(file)
input.setEncoding("utf-8")
input.setCharacterStream(file)
# There is a bug in xml.sax.saxutils.prepare_input_source
input.setByteStream(file)
input.setSystemId(None)
elif (isinstance(self.source,InputSource)):
input = self.source
else:
raise Exception("Parse source must be either string or InputSource")
# Create the parser/xmlreader
parser = xml.sax.make_parser()
# Tell the parser to use our handler(s)
parser.setContentHandler(self)
#parser.setErrorHandler(self)
#parser.setFeature(xml.sax.handler.feature_namespaces,True)
# Shut off dtd validation
parser.setFeature(xml.sax.handler.feature_validation,False)
parser.setFeature(xml.sax.handler.feature_external_ges, False)
# Parse the document
parser.parse(input)
def test_parse_InputSource(self):
# accept data without declared but with explicitly specified encoding
make_xml_file(self.data, 'iso-8859-1', None)
with open(TESTFN, 'rb') as f:
input = InputSource()
input.setByteStream(f)
input.setEncoding('iso-8859-1')
self.check_parse(input)
def test_parse_InputSource(self):
# accept data without declared but with explicitly specified encoding
make_xml_file(self.data, 'iso-8859-1', None)
with open(TESTFN, 'rb') as f:
input = InputSource()
input.setByteStream(f)
input.setEncoding('iso-8859-1')
self.check_parse(input)
def parse(self, charSet, file):
'''
Parses the provided content.
@param charSet: string
The character set of the content.
@param file: byte file
The bytes file object providing the content.
@return: object
The object obtained from parsing.
'''
inpsrc = InputSource()
inpsrc.setByteStream(file)
inpsrc.setEncoding(charSet)
return self.parseInputSource(inpsrc)
def parse(self, charSet, file):
'''
Parses the provided content.
@param charSet: string
The character set of the content.
@param file: byte file
The bytes file object providing the content.
@return: object
The object obtained from parsing.
'''
inpsrc = InputSource()
inpsrc.setByteStream(file)
inpsrc.setEncoding(charSet)
return self.parseInputSource(inpsrc)
def parse(self, source, charSet, decoding, target):
'''
@see: ParseBaseHandler.parse
'''
assert isinstance(source, IInputStream), 'Invalid stream %s' % source
assert isinstance(charSet, str), 'Invalid character set %s' % charSet
assert isinstance(target, Target), 'Invalid target %s' % target
parse = Parse(self.parser, decoding, target)
self.parser.setContentHandler(parse)
inpsrc = InputSource()
inpsrc.setByteStream(source)
inpsrc.setEncoding(charSet)
try: self.parser.parse(source)
except SAXParseException as e:
assert isinstance(e, SAXParseException)
addFailure(target, decoding, 'Bad XML content at line %(line)s and column %(column)s',
line=e.getLineNumber(), column=e.getColumnNumber())
def parse(self, decoder, data, source, charSet):
'''
@see: ParseBaseHandler.parse
'''
assert callable(decoder), 'Invalid decoder %s' % decoder
assert isinstance(data, dict), 'Invalid data %s' % data
assert isinstance(source, IInputStream), 'Invalid stream %s' % source
assert isinstance(charSet, str), 'Invalid character set %s' % charSet
parse = Parse(self.parser, decoder, data)
self.parser.setContentHandler(parse)
inpsrc = InputSource()
inpsrc.setByteStream(source)
inpsrc.setEncoding(charSet)
try: self.parser.parse(source)
except SAXParseException as e:
assert isinstance(e, SAXParseException)
return 'Bad XML content at line %s and column %s' % (e.getLineNumber(), e.getColumnNumber())
except ParseError as e:
assert isinstance(e, ParseError)
return str(e)
def parse(self, source, charSet, decoding, target):
'''
@see: ParseBaseHandler.parse
'''
assert isinstance(source, IInputStream), 'Invalid stream %s' % source
assert isinstance(charSet, str), 'Invalid character set %s' % charSet
assert isinstance(target, Target), 'Invalid target %s' % target
parse = Parse(self.parser, decoding, target)
self.parser.setContentHandler(parse)
inpsrc = InputSource()
inpsrc.setByteStream(source)
inpsrc.setEncoding(charSet)
try:
self.parser.parse(source)
except SAXParseException as e:
assert isinstance(e, SAXParseException)
addFailure(
target,
decoding,
'Bad XML content at line %(line)s and column %(column)s',
line=e.getLineNumber(),
column=e.getColumnNumber())
