def test_parse_again(self):
parser = expat.ParserCreate()
file = StringIO.StringIO(data)
parser.ParseFile(file)
# Issue 6676: ensure a meaningful exception is raised when attempting
# to parse more than one XML document per xmlparser instance,
# a limitation of the Expat library.
with self.assertRaises(expat.error) as cm:
parser.ParseFile(file)
self.assertEqual(expat.ErrorString(cm.exception.code),
expat.errors.XML_ERROR_FINISHED)
def parse(self, input, root=None, bInRootTag=0):
""" Builder.parse """
# prepare builder
self._unknownTagName = None
self.oRoot = root
self.oRootNode = None
self.oCurrNode = None
self.bInRootTag = bInRootTag
self.bInCData = False
if bInRootTag:
self.oCurrNode = root
# create parser object
p = pyexpat.ParserCreate()
# connect parser object with handler methods
p.StartElementHandler = self.OnStartElement
p.EndElementHandler = self.OnEndElement
p.CharacterDataHandler = self.OnCharacterData
p.StartCdataSectionHandler = self.OnStartCData
p.EndCdataSectionHandler = self.OnEndCData
p.ProcessingInstructionHandler = self.OnProcessingInstruction
p.CommentHandler = self.OnComment
p.StartNamespaceDeclHandler = self.OnStartNamespaceDecl
p.EndNamespaceDeclHandler = self.OnEndNamespaceDecl
#### parsing ####
_globals.writeLog( self, "#### parsing ####")
if type(input) is str:
# input is a string!
rv = p.Parse(input, 1)
else:
# input is a file object!
while True:
v=input.read(self.iBufferSize)
if v=="":
rv = 1
break
rv = p.Parse(v, 0)
if not rv:
break
# raise parser exception
if not rv:
raise ParseError('%s at line %s' % (pyexpat.ErrorString(p.ErrorCode), p.ErrorLineNumber))
####
return self.oRootNode
def parse(self, input, root=None, bInRootTag=0):
""" Builder.parse """
# prepare builder
self.oRootTag = None
self.oCurrNode = None
self.bInCData = False
if bInRootTag:
self.oCurrNode = root
# create parser object
p = pyexpat.ParserCreate()
# connect parser object with handler methods
p.StartElementHandler = self.OnStartElement
p.EndElementHandler = self.OnEndElement
p.CharacterDataHandler = self.OnCharacterData
p.StartCdataSectionHandler = self.OnStartCData
p.EndCdataSectionHandler = self.OnEndCData
p.ProcessingInstructionHandler = self.OnProcessingInstruction
p.CommentHandler = self.OnComment
p.StartNamespaceDeclHandler = self.OnStartNamespaceDecl
p.EndNamespaceDeclHandler = self.OnEndNamespaceDecl
#### parsing ####
if isinstance(input,bytes):
# input is a string!
rv = p.Parse(input, 1)
else:
# input is a file object!
while True:
v=input.read(self.iBufferSize)
if v=="":
rv = 1
break
rv = p.Parse(v, 0)
if not rv:
break
# raise parser exception
if not rv:
raise ParseError('%s at line %s' % (pyexpat.ErrorString(p.ErrorCode), p.ErrorLineNumber))
####
# reindex
standard.triggerEvent( self, '*.ObjectImported')
####
return root
def test_build():
def entry_point(argv):
res = interp_pyexpat.XML_ErrorString(3)
os.write(1, rffi.charp2str(res))
return 0
t = TranslationContext()
t.buildannotator().build_types(entry_point, [s_list_of_strings])
t.buildrtyper().specialize()
builder = CStandaloneBuilder(t, entry_point, t.config)
builder.generate_source()
builder.compile()
data = builder.cmdexec()
assert data == pyexpat.ErrorString(3)
def parse(self, input, mediadbStorable=True):
""" XmlAttrBuilder.parse """
# prepare builder
self.mediadbStorable = mediadbStorable
self.dValueStack = _globals.MyStack()
self.dTagStack = _globals.MyStack()
# create parser object
p = pyexpat.ParserCreate()
# connect parser object with handler methods
p.StartElementHandler = self.OnStartElement
p.EndElementHandler = self.OnEndElement
p.CharacterDataHandler = self.OnCharacterData
p.StartCdataSectionHandler = self.OnStartCData
p.EndCdataSectionHandler = self.OnEndCData
p.ProcessingInstructionHandler = self.OnProcessingInstruction
p.CommentHandler = self.OnComment
p.StartNamespaceDeclHandler = self.OnStartNamespaceDecl
p.EndNamespaceDeclHandler = self.OnEndNamespaceDecl
#### parsing ####
if type(input) is str:
# input is a string!
rv = p.Parse(input, 1)
else:
# input is a file object!
while True:
v = input.read(self.iBufferSize)
if v == "":
rv = 1
break
rv = p.Parse(v, 0)
if not rv:
break
# raise parser exception
if not rv:
raise ParseError(
'%s at line %s' %
(pyexpat.ErrorString(p.ErrorCode), p.ErrorLineNumber))
return self.dValueStack.pop()
def parse(self, input):
""" XmlBuilder.parse """
# prepare builder
self.dTagStack = collections.deque()
self.dTagStack.append({'tags': []})
# create parser object
p = pyexpat.ParserCreate()
# connect parser object with handler methods
p.StartElementHandler = self.OnStartElement
p.EndElementHandler = self.OnEndElement
p.CharacterDataHandler = self.OnCharacterData
p.StartCdataSectionHandler = self.OnStartCData
p.EndCdataSectionHandler = self.OnEndCData
p.ProcessingInstructionHandler = self.OnProcessingInstruction
p.CommentHandler = self.OnComment
p.StartNamespaceDeclHandler = self.OnStartNamespaceDecl
p.EndNamespaceDeclHandler = self.OnEndNamespaceDecl
#### parsing ####
if isinstance(input, bytes):
# input is a string!
rv = p.Parse(input, 1)
else:
# input is a file object!
while True:
v = input.read(self.iBufferSize)
if len(v) == 0:
rv = 1
break
rv = p.Parse(v, 0)
if not rv:
break
# raise parser exception
if not rv:
raise ParseError(
'%s at line %s' %
(pyexpat.ErrorString(p.ErrorCode), p.ErrorLineNumber))
return self.dTagStack.pop()['tags']
def test_ErrorString(self):
self.assertEqual(expat.ErrorString(1), 'out of memory')
data = """<?xml version="1.0" encoding="iso-8859-1" standalone="no"?>
<?xml-stylesheet href="stylesheet.css"?>
<!-- comment data -->
<!DOCTYPE quotations SYSTEM "quotations.dtd" [
<!ELEMENT root ANY>
<!NOTATION notation SYSTEM "notation.jpeg">
<!ENTITY acirc "â">
<!ENTITY external_entity SYSTEM "entity.file">
<!ENTITY unparsed_entity SYSTEM "entity.file" NDATA notation>
%unparsed_entity;
]>
<root>
<myns:subelement xmlns:myns="http://www.python.org/namespace">
Contents of subelements
</myns:subelement>
<sub2><![CDATA[contents of CDATA section]]></sub2>
&external_entity;
</root>
"""
try:
parser.Parse(data, 1)
except pyexpat.error:
print '** Error', parser.ErrorCode, pyexpat.ErrorString( parser.ErrorCode)
print '** Line', parser.ErrorLineNumber
print '** Column', parser.ErrorColumnNumber
print '** Byte', parser.ErrorByteIndex
