def _handleLxmlConfigNode(self, session, node):
if (node.tag == "cluster"):
maps = []
for child in node.iterchildren(tag=etree.Element):
if (child.tag == "map"):
t = child.attrib.get('type', '')
map = []
for xpchild in child.iterchildren(tag=etree.Element):
if (xpchild.tag == "xpath"):
map.append(flattenTexts(xpchild).strip())
elif (xpchild.tag == "process"):
# turn xpath chain to workflow
ref = xpchild.attrib.get('ref', None)
if ref is not None:
process = self.get_object(session, ref)
else:
xpchild.tag = 'workflow'
process = CachingWorkflow(
session, xpchild, self)
process._handleLxmlConfigNode(session, xpchild)
map.append(process)
#vxp = [map[0]]
if (len(map) < 3):
# default ExactExtractor
map.append([['extractor', 'SimpleExtractor']])
if (t == u'key'):
self.keyMap = [map[0], map[1], map[2]]
else:
maps.append([map[0], map[1], map[2]])
self.maps = maps
def _handleLxmlConfigNode(self, session, node):
# Source
if (node.tag == "source"):
key = node.attrib.get('id', None)
default = node.attrib.get('default', None)
process = None
preprocess = None
xp = None
for child in node.iterchildren(tag=etree.Element):
if child.tag == "xpath":
if xp == None:
ref = child.attrib.get('ref', '')
if ref:
xp = self.get_object(session, ref)
else:
node.set('id', self.id + '-xpath')
xp = SimpleXPathProcessor(session, node, self)
xp._handleLxmlConfigNode(session, node)
elif child.tag == "preprocess":
# turn preprocess chain to workflow
ref = child.attrib.get('ref', '')
if ref:
preprocess = self.get_object(session, ref)
else:
# create new element
e = etree.XML(etree.tostring(child))
e.tag = 'workflow'
e.set('id', self.id + "-preworkflow")
preprocess = CachingWorkflow(session, child, self)
preprocess._handleLxmlConfigNode(session, child)
elif child.tag == "process":
# turn xpath chain to workflow
ref = child.attrib.get('ref', '')
if ref:
process = self.get_object(session, ref)
else:
# create new element
e = etree.XML(etree.tostring(child))
e.tag = 'workflow'
e.set('id', self.id + "-workflow")
process = CachingWorkflow(session, e, self)
process._handleLxmlConfigNode(session, e)
self.sources[key] = {
'source': (xp, process, preprocess),
'default': default
}
def _handleConfigNode(self, session, node):
if (node.localName == "cluster"):
maps = []
for child in node.childNodes:
if (child.nodeType == elementType
and child.localName == "map"):
t = child.getAttributeNS(None, 'type')
map = []
for xpchild in child.childNodes:
if (xpchild.nodeType == elementType
and xpchild.localName == "xpath"):
map.append(flattenTexts(xpchild))
elif (xpchild.nodeType == elementType
and xpchild.localName == "process"):
# turn xpath chain to workflow
ref = xpchild.getAttributeNS(None, 'ref')
if ref:
process = self.get_object(session, ref)
else:
try:
xpchild.localName = 'workflow'
except:
# 4suite dom sets read only
newTop = xpchild.ownerDocument.createElementNS(
None, 'workflow')
for kid in xpchild.childNodes:
newTop.appendChild(kid)
xpchild = newTop
process = CachingWorkflow(
session, xpchild, self)
process._handleConfigNode(session, xpchild)
map.append(process)
# XXX FIX ME
# vxp = verifyXPaths([map[0]])
vxp = [map[0]]
if (len(map) < 3):
# default ExactExtractor
map.append([['extractor', 'SimpleExtractor']])
if (t == u'key'):
self.keyMap = [vxp[0], map[1], map[2]]
else:
maps.append([vxp[0], map[1], map[2]])
self.maps = maps
