class NeuralEditNeuralNet(PickleToXML):
__pickle_to_xml__ = ['NetPath', 'Elements']
# The idea with pickling this is to save a REFERENCE to the Net file
# and pickle the Net in its own file - the editor handles
# saving and restoring the net as well as re-establishing the self.Net property
#
def __init__(self):
'''
A container for NeuralEditElement representing the GUI components of a neural net
'''
self.Net = NeuralNet()
self.NetPath = None # set during pickle op
self.Elements = [] # elements are UI representation of individual neurons
self.LookupTable = {}
def to_json(self):
''' convert UI Net to json repr
{ 'nodes': { 'node_x' : {'x': float, 'y': float }, ... }, 'edges': [[node_source, node_target], ... ] }
'''
json = {'nodes': {}, 'edges': [], 'outputs': [], 'inputs': []}
for e in self.Elements:
json['nodes'][e.Name] = {'x': e.Position[0], 'y': e.Position[1]}
# add edges, outputs, inputs
neuron = self.lookup_neuron(e)
if len(neuron.Outgoing) == 0:
if len(neuron.Incoming) != 0:
json['outputs'].append(e.Name)
else:
if len(neuron.Incoming) == 0:
json['inputs'].append(e.Name)
# record edges on the outgoing side
for link in neuron.Outgoing:
target = self.lookup_element(link.Target)
json['edges'].append([e.Name, target.Name])
# add outputs
# add inputs
return json
def element_from_point(self, point):
for e in reversed(self.Elements): # reverse order of hit test from painting
if e.hit_test(point):
return e
return None
def rebuild_lookup_table(self):
self.LookupTable = {}
for e in self.Elements:
self.LookupTable[e.Name] = (self.Net.lookup_neuron_by_name(e.Name), e)
def lookup_neuron(self, element):
return None if element is None else self.LookupTable[element.Name][0]
def lookup_element(self, neuron):
return None if neuron is None else self.LookupTable[neuron.Name][1]
def add_named_element(self, position, name):
neuron = self.Net.add_neuron()
neuron.Name = name
element = NeuralEditElement(name, position)
self.Elements.append(element)
self.LookupTable[name] = (neuron, element)
return element
def add_element(self, position, neuron_type, path):
if path is None:
neuron = self.Net.add_neuron()
else:
neuron = self.Net.add_subnet(neuron_type, path)
element = NeuralEditElement(neuron.Name, position)
self.Elements.append(element)
self.LookupTable[neuron.Name] = (neuron, element)
return element
def add_link(self, start_element, end_element):
return self.Net.add_link(self.lookup_neuron(start_element), self.lookup_neuron(end_element))
def remove_element(self, element):
(neuron, element) = self.LookupTable[element.Name]
self.Elements.remove(element)
self.Net.remove_neuron(neuron)
self.LookupTable.pop(element.Name)
def rename_element(self, element, name):
(neuron, element) = self.LookupTable[element.Name]
self.LookupTable.pop(element.Name)
neuron.Name = "" # temp to let get_unique_name reuse our name
name = self.Net.get_unique_name(name)
neuron.Name = name
element.Name = name
self.LookupTable[neuron.Name] = (neuron, element)
return name # in case the selected name is different from what was passed in
