def filterOac(nodes):
result = []
for node in nodes:
namespace, name = splitType(getAttrib(node, 'rdf:resource'))
if namespace == namespaces['oa']:
result.append(node)
return result
def process(self, lxmlNode):
def prefixForNamespace(ns):
if ns in prefixes:
return prefixes[ns]
return self._namespaceEnum.get(ns)
def filterOac(nodes):
result = []
for node in nodes:
namespace, name = splitType(getAttrib(node, 'rdf:resource'))
if namespace == namespaces['oa']:
result.append(node)
return result
descriptions = xpath(lxmlNode, "//rdf:Description[rdf:type]")
for description in descriptions:
nodes = xpath(description, "rdf:type")
oacNodes = filterOac(nodes)
typeNode = nodes[0]
if len(oacNodes) > 0:
typeNode = oacNodes[0]
parent = description.getparent()
namespace, name = splitType(getAttrib(typeNode, 'rdf:resource'))
prefix = prefixForNamespace(namespace)
newNode = SubElement(
parent,
"{%(ns)s}%(tag)s" % {'ns': namespace, 'tag': name},
attrib=description.attrib,
nsmap={prefix: namespace})
for child in (child for child in description.getchildren() if child != typeNode):
newNode.append(child)
parent.remove(description)
yield self.all.process(lxmlNode=lxmlNode)
def process(self, lxmlNode):
for node in chain(xpath(lxmlNode, '//oa:Annotation'), xpath(lxmlNode, '//oa:Body'), xpath(lxmlNode, "//oa:SpecificResource")):
if getAttrib(node, 'rdf:about') == None and getAttrib(node, 'rdf:resource') == None:
setAttrib(node, 'rdf:about', self._urnGen())
yield self.all.process(lxmlNode=lxmlNode)
