def _set_coordinator_properties(coordinator, root, namespace):
"""
Get coordinator properties from coordinator XML
Set properties on ``coordinator`` with attributes from XML etree ``root``.
"""
coordinator.name = root.get('name')
coordinator.timezone = root.get('timezone')
coordinator.start = oozie_to_django_datetime(root.get('start'))
coordinator.end = oozie_to_django_datetime(root.get('end'))
coordinator.frequency_unit, coordinator.frequency_number = oozie_to_hue_frequency(root.get('frequency'))
def _set_coordinator_properties(coordinator, root, namespace):
"""
Get coordinator properties from coordinator XML
Set properties on ``coordinator`` with attributes from XML etree ``root``.
"""
coordinator.name = root.get('name')
coordinator.timezone = root.get('timezone')
coordinator.start = oozie_to_django_datetime(root.get('start'))
coordinator.end = oozie_to_django_datetime(root.get('end'))
coordinator.frequency_unit, coordinator.frequency_number = oozie_to_hue_frequency(
root.get('frequency'))
def _set_controls(bundle, root, namespace):
"""
Get controls from bundle XML
Set properties on ``bundle`` with controls from XML etree ``root``.
"""
namespaces = {'n': namespace}
controls = root.xpath('n:controls', namespaces=namespaces)[0]
kick_off_time = controls.xpath('n:kick-off-time', namespaces=namespaces)
if kick_off_time:
bundle.kick_off_time = oozie_to_django_datetime(kick_off_time[0].text)
def _reconcile_datasets(coordinator, objects, root, namespace):
"""
Reconcile datasets, datainputs, and dataoutputs.
datasets and tied to datainputs and dataoutputs.
Other dataset properties are set from XML tree.
``coordinator`` is used for look ups and creating relevant objects.
``objects`` should be transformed from XSLTs and loaded via django deserializers.
``root`` is the element root of the coordinator (lxml).
``namespace`` defines the XML coordinator namespace and schema version.
"""
namespaces = {'n': namespace}
datasets = {}
datainputs = []
dataoutputs = []
coordinator.save()
for obj in objects:
obj.object.coordinator = coordinator
if isinstance(obj.object, Dataset):
obj.object.pk = None
obj.object.save()
dataset_elements = root.xpath('//n:dataset[@name="%s"]' %
obj.object.name,
namespaces=namespaces)
for dataset_element in dataset_elements:
obj.object.start = oozie_to_django_datetime(
dataset_element.get('initial-instance'))
datasets[obj.object.name] = obj.object
obj.object.save()
elif isinstance(obj.object, DataInput):
datainputs.append(obj.object)
elif isinstance(obj.object, DataOutput):
dataoutputs.append(obj.object)
for datainput in datainputs:
datainput_elements = root.xpath('//n:data-in[@name="%s"]' %
datainput.name,
namespaces=namespaces)
for datainput_element in datainput_elements:
datainput.dataset = datasets[datainput_element.get('dataset')]
datainput.pk = None
datainput.save()
for dataoutput in dataoutputs:
dataoutput_elements = root.xpath('//n:data-out[@name="%s"]' %
dataoutput.name,
namespaces=namespaces)
for dataoutput_element in dataoutput_elements:
dataoutput.dataset = datasets[dataoutput_element.get('dataset')]
dataoutput.pk = None
dataoutput.save()
def _set_controls(bundle, root, namespace):
"""
Get controls from bundle XML
Set properties on ``bundle`` with controls from XML etree ``root``.
"""
namespaces = {
'n': namespace
}
controls = root.xpath('n:controls', namespaces=namespaces)[0]
kick_off_time = controls.xpath('n:kick-off-time', namespaces=namespaces)
if kick_off_time:
bundle.kick_off_time = oozie_to_django_datetime(kick_off_time[0].text)
def _reconcile_datasets(coordinator, objects, root, namespace):
"""
Reconcile datasets, datainputs, and dataoutputs.
datasets and tied to datainputs and dataoutputs.
Other dataset properties are set from XML tree.
``coordinator`` is used for look ups and creating relevant objects.
``objects`` should be transformed from XSLTs and loaded via django deserializers.
``root`` is the element root of the coordinator (lxml).
``namespace`` defines the XML coordinator namespace and schema version.
"""
namespaces = {
'n': namespace
}
datasets = {}
datainputs = []
dataoutputs = []
coordinator.save()
for obj in objects:
obj.object.coordinator = coordinator
if isinstance(obj.object, Dataset):
obj.object.pk = None
obj.object.save()
dataset_elements = root.xpath('//n:dataset[@name="%s"]' % obj.object.name, namespaces=namespaces)
for dataset_element in dataset_elements:
obj.object.start = oozie_to_django_datetime(dataset_element.get('initial-instance'))
datasets[obj.object.name] = obj.object
obj.object.save()
elif isinstance(obj.object, DataInput):
datainputs.append(obj.object)
elif isinstance(obj.object, DataOutput):
dataoutputs.append(obj.object)
for datainput in datainputs:
datainput_elements = root.xpath('//n:data-in[@name="%s"]' % datainput.name, namespaces=namespaces)
for datainput_element in datainput_elements:
datainput.dataset = datasets[datainput_element.get('dataset')]
datainput.pk = None
datainput.save()
for dataoutput in dataoutputs:
dataoutput_elements = root.xpath('//n:data-out[@name="%s"]' % dataoutput.name, namespaces=namespaces)
for dataoutput_element in dataoutput_elements:
dataoutput.dataset = datasets[dataoutput_element.get('dataset')]
dataoutput.pk = None
dataoutput.save()