def validateTask(self, taskConf, taskArgumentDefinition):
"""
_validateTask_
Validate the task information against the given
argument description
"""
try:
validateArgumentsCreate(taskConf,
taskArgumentDefinition,
checkInputDset=False)
# Validate GPU-related spec parameters
StdBase.validateGPUSettings(taskConf)
except WMSpecFactoryException:
# just re-raise it to keep the error message clear
raise
except Exception as ex:
self.raiseValidationException(str(ex))
def testValidateGPUSettings(self):
"""
Test the 'validateGPUSettings' StdBase method.
"""
with self.assertRaises(WMSpecFactoryException):
StdBase.validateGPUSettings({"RequiresGPU": "optional"})
with self.assertRaises(WMSpecFactoryException):
StdBase.validateGPUSettings({"RequiresGPU": "required"})
with self.assertRaises(WMSpecFactoryException):
StdBase.validateGPUSettings({
"RequiresGPU": "optional",
"GPUParams": json.dumps("")
})
with self.assertRaises(WMSpecFactoryException):
StdBase.validateGPUSettings({
"RequiresGPU": "required",
"GPUParams": json.dumps(None)
})
# now input that passes the validation
self.assertTrue(
StdBase.validateGPUSettings({"RequiresGPU": "forbidden"}))
self.assertTrue(
StdBase.validateGPUSettings({
"RequiresGPU": "optional",
"GPUParams": json.dumps("blah")
}))
self.assertTrue(
StdBase.validateGPUSettings({
"RequiresGPU":
"required",
"GPUParams":
json.dumps({"Key1": "value1"})
}))
def validateSchema(self, schema):
"""
_validateSchema_
Go over each task and make sure it matches validation
parameters derived from Dave's requirements. They are:
* cannot have more than 10 tasks
* output from the last task *must* be saved
* task must be described as a python dictionary type
* transient output modules must be an input for further task(s)
* and the usual Task arguments validation, as defined in the spec
"""
numTasks = schema['TaskChain']
if numTasks > 10:
msg = "Workflow exceeds the maximum allowed number of tasks. "
msg += "Limited to up to 10 tasks, found %s tasks." % numTasks
self.raiseValidationException(msg)
lastTask = "Task%s" % schema['TaskChain']
if not strToBool(schema[lastTask].get('KeepOutput', True)):
msg = "Dropping the output (KeepOutput=False) of the last task is prohibited.\n"
msg += "You probably want to remove that task completely and try again."
self.raiseValidationException(msg=msg)
transientMapping = {}
for i in range(1, numTasks + 1):
taskNumber = "Task%s" % i
if taskNumber not in schema:
msg = "Task '%s' not present in the request schema." % taskNumber
self.raiseValidationException(msg=msg)
task = schema[taskNumber]
# We can't handle non-dictionary tasks
if not isinstance(task, dict):
msg = "Task '%s' not defined as a dictionary. " % taskNumber
msg += "It could be an indicator of JSON error.\n"
self.raiseValidationException(msg=msg)
# Generic task parameter validation
self.validateTask(
task,
self.getChainCreateArgs(i == 1, i == 1
and 'InputDataset' not in task))
# Validate the existence of the configCache
if task["ConfigCacheID"]:
self.validateConfigCacheExists(
configID=task['ConfigCacheID'],
configCacheUrl=schema["ConfigCacheUrl"],
couchDBName=schema["CouchDBName"],
getOutputModules=False)
# Validate the chaining of transient output modules, need to make a copy of the lists
transientMapping[task['TaskName']] = [
x for x in task.get('TransientOutputModules', [])
]
if i > 1:
inputTransientModules = transientMapping[task['InputTask']]
if task['InputFromOutputModule'] in inputTransientModules:
inputTransientModules.remove(task['InputFromOutputModule'])
try:
StdBase.validateGPUSettings(schema)
except Exception as ex:
self.raiseValidationException(str(ex))
for task in transientMapping:
if transientMapping[task]:
msg = "A transient module is not processed by a subsequent task.\n"
msg += "This is a malformed task chain workload"
self.raiseValidationException(msg)
