def POST(self, context_classes):
print("--- InitClassifier POST request ---")
classes_list = json.loads(context_classes)
model_exists_result = check_model_exists(classes_list)
if model_exists_result == False:
# This model does not exists on server yet, so we have to build it:
# Create random ID for this request:
id = str(randint(1e5,1e6))
# Create an id, where the client can get the new classifier when the calculation is done:
filename_new_classifier = str(id) + ".json"
valid_classes = []
invalid_classes = []
waitOrNoWait = "no_wait"
for el in classes_list:
if (feasibilityCheck(el) == "not_feasible"):
invalid_classes.append(el)
elif (feasibilityCheck(el) == "downloaded"):
valid_classes.append(el)
elif (feasibilityCheck(el) == "feasible"):
valid_classes.append(el)
else:
print("feasibilityCheck returned invalid result")
if len(valid_classes) != 0:
# Start training the classifier in the background (list is passed as String and converted back with Regex later):
subprocess.Popen(["python", "server_create_initial_model.py", json.dumps(valid_classes), filename_new_classifier])
waitOrNoWait = "wait"
else:
# If not a single class is valid, don't do anything:
waitOrNoWait = "no_wait"
# Return new filename and the list of invalid classes:
response = {"filename": filename_new_classifier, "wait": waitOrNoWait, "invalid_classes": invalid_classes}
else:
# The model already exisits and the location of the model can be send to server:
filename_exisiting = "classifiers/" + model_exists_result
#response = {"wait": "no_wait", "filename": filename_exisiting}
response = {"filename": filename_exisiting, "wait": "no_wait", "invalid_classes": []}
return response
def POST(self, classname):
print("--- FeasibilityCheck POST request ---")
res = feasibilityCheck(classname)
return res
def POST(self, context_classes):
print("--- ManageClasses POST request ---")
context_classes_list = json.loads(context_classes)
classifier_json = cherrypy.request.json
# Create random ID for this request:
id = str(randint(1e5,1e6))
# Dump the GMM object we received from the client, as it is too large to parse as argument directly:
filename_old_classifier = "tmp_" + id + ".json"
json.dump(classifier_json, open(filename_old_classifier,"wb"))
# Create an id, where the client can get the new classifier when the calculation is done:
filename_new_classifier = "Classifier_" + id + ".json"
valid_classes = []
invalid_classes = []
for el in context_classes_list:
if (feasibilityCheck(el) == "not_feasible"):
invalid_classes.append(el)
waitOrNoWait = "no_wait"
elif (feasibilityCheck(el) == "downloaded"):
valid_classes.append(el)
waitOrNoWait = "no_wait"
elif (feasibilityCheck(el) == "feasible"):
valid_classes.append(el)
waitOrNoWait = "wait"
else:
print("feasibilityCheck returned invalid result")
waitOrNoWait = "no_wait"
# Start training the classifier in the background:
subprocess.Popen(["python", "server_manage_classes.py", json.dumps(valid_classes), filename_old_classifier, filename_new_classifier])
# Return new filename and the list of invalid classes:
response = {"filename": filename_new_classifier, "wait": waitOrNoWait, "invalid_classes": invalid_classes}
return response
