KeyError in Edge Categorical and Edge Numerical Compartments

[neserende]

Describe the bug
When working with a random regular graph generated by networkx, sometimes edges are listed as (large number, small number). Example: (30, 5) instead of (5, 30). This type of data can also come when network is read from a file input. In the edge categorical attribute and edge numerical attribute compartments, a key error occurs when this case happens.

In edge categorical attribute:

def execute(self, node, graph, status, status_map, *args, **kwargs):
        neighbors = list(graph.neighbors(node))
        if isinstance(graph, nx.DiGraph):
            neighbors = list(graph.predecessors(node))

        edge_attr = graph.get_edge_attributes(self.attribute)

        if self.trigger is not None:
            triggered = [v for v in neighbors if status[v] == status_map[self.trigger] and
                         edge_attr[(min([node, v]), max([node, v]))] == self.attribute_value]

Here, edge_attr holds a category ('co-worker') for (30, 5) but edge_attr[(min([node, v]), max([node, v]))] tries to reach (5, 30) which does not exist and results in the following key error:

File "C:\Users\NESE\AppData\Local\Packages\PythonSoftwareFoundation.Python.3.9_qbz5n2kfra8p0\LocalCache\local-packages\Python39\site-packages\ndlib\models\compartments\EdgeCategoricalAttribute.py", line 32, in
edge_attr[(min([node, v]), max([node, v]))] == self.attribute_value]
KeyError: (5, 30)

In node categorical attribute:

    def execute(self, node, graph, status, status_map, *args, **kwargs):

        neighbors = list(graph.neighbors(node))
        if isinstance(graph, nx.DiGraph):
            neighbors = list(graph.predecessors(node))

        edge_attr = graph.get_edge_attributes(self.attribute)

        triggered = []

        if self.trigger is not None:
            for v in neighbors:
                if status[v] == status_map[self.trigger]:
                    val = edge_attr[(min([node, v]), max([node, v]))]
                    if self.operator == "IN":
                        if self.__available_operators[self.operator][0](val, self.attribute_range[0]) and \
                                    self.__available_operators[self.operator][1](val, self.attribute_range[1]):
                            triggered.append(v)
                    else:
                        if self.__available_operators[self.operator](val, self.attribute_range):
                            triggered.append(v)
        else:
            for v in neighbors:
                val = edge_attr[(min([node, v]), max([node, v]))]

Here, edge_attr dictionary is accessed in a similar manner and a key error occurs again:

File "C:\Users\NESE\AppData\Local\Packages\PythonSoftwareFoundation.Python.3.9_qbz5n2kfra8p0\LocalCache\local-packages\Python39\site-packages\ndlib\models\compartments\EdgeNumericalAttribute.py", line 63, in execute
val = edge_attr[(min([node, v]), max([node, v]))]
KeyError: (23, 68)

To Reproduce
Steps to reproduce the behavior:

• NDlib version: v5.1.1
• Operating System: Windows
• Python version: 3.9
• Version(s) of NDlib required libraries: Version installed with ndlib v5.1.1

Expected behavior
Before accessing edge_attr dictionary, if (5, 30) key does not exist, we should access (30, 5) instead.

Additional context
The code I used to test edge categorical compartments:

import random
import ndlib.models.CompositeModel as gc
import ndlib.models.compartments as cpm
import networkx as nx
import ndlib.models.ModelConfig as mc


# Network generation
#A regular graph is a graph where each node has the same number of neighbors.
#Here we create a random graph with 100 nodes where each node has a degree of 4
g = nx.random_regular_graph(4, 100) 

# Setting edge attribute
attr = {(u, v): {"type": random.choice(['co-worker', 'family'])} for (u, v) in g.edges()}
nx.set_edge_attributes(g, attr)

# Composite Model instantiation
model = gc.CompositeModel(g)

# Model statuses
model.add_status("Susceptible")
model.add_status("Infected")
model.add_status("Removed")

# Compartment definition
c1 = cpm.EdgeCategoricalAttribute("type", "family", probability = 0.06, triggering_status="Susceptible")

# Rule definition
model.add_rule("Susceptible", "Infected", c1)

# Model initial status configurations
config = mc.Configuration()
config.add_model_parameter('percentage_infected', 0.1)

# Simulation execution
model.set_initial_status(config)
iterations = model.iteration_bunch(10)

trends = model.build_trends(iterations)

print(trends)

The code I used to test edge numerical compartments:

import random
import ndlib.models.CompositeModel as gc
import ndlib.models.compartments as cpm
import networkx as nx
import ndlib.models.ModelConfig as mc


# Network generation
#A regular graph is a graph where each node has the same number of neighbors.
#Here we create a random graph with 100 nodes where each node has a degree of 4
g = nx.random_regular_graph(4, 100) 

# Setting edge attribute
attr = {(u, v): {"weight": int((u+v) % 10)} for (u, v) in g.edges()}
nx.set_edge_attributes(g, attr)

# Composite Model instantiation
model = gc.CompositeModel(g)

# Model statuses
model.add_status("Susceptible")
model.add_status("Infected")
model.add_status("Removed")

# Compartment definition
c1 = cpm.EdgeNumericalAttribute("weight", value=4, op="==", probability=0.6)

# Rule definition
model.add_rule("Susceptible", "Infected", c1)

# Model initial status configurations
config = mc.Configuration()
config.add_model_parameter('percentage_infected', 0.1)

# Simulation execution
model.set_initial_status(config)
iterations = model.iteration_bunch(10)

trends = model.build_trends(iterations)

print(trends)