NetworkX - allow duplicate nodes - python-3.x

Question

import networkx as nx
import matplotlib.pyplot as plt
G = nx.DiGraph()
def calculate_lists(user_input):
""" Calculates the number of occurences of certain character in a string."""
input_list = []
for i in user_input:
input_list.append(i)
occurence_list = []
for i in set(input_list):
occurence_list.append((i, user_input.count(i)))
sorted_by_first = sorted(occurence_list, key=lambda tup: tup[1])
sorted_list = list(reversed(sorted_by_first))
propability_list = []
for i in range(len(sorted_list)):
propability_list.append(sorted_list[i][1])
print("Input list is: ", input_list)
print("Input list is: ", input_list)
print("Occurence list: ", occurence_list)
print("Sorted list is: ", sorted_list)
print("Probility list is: ", propability_list)
return huffmann_algorithm(propability_list)
def huffmann_algorithm(prob_list):
node_list = []
while len(prob_list) != 1:
first_minimum = min(float(s) for s in prob_list)
print("First minimum", first_minimum)
prob_list.remove(first_minimum)
second_minimum = min(float(s) for s in prob_list)
print("Second minimum", second_minimum)
prob_list.remove(second_minimum)
node_list.append([first_minimum, second_minimum])
print("new value: ", first_minimum+second_minimum)
new_value = int(first_minimum+second_minimum)
prob_list.append(new_value)
print("Finished: ", prob_list)
count = 0
for i in node_list:
print(count)
print("Nodes: ", tuple(i))
G.add_node(i[0])
G.add_node(i[1])
G.add_node(i[0]+i[1])
G.add_edge(i[0], i[0]+i[1])
G.add_edge(i[1], i[0]+i[1])
print("Node list: ", node_list)
print(G.nodes())
nx.draw_networkx(G, with_labels=True, arrows=False)
plt.savefig("graph1.png")
plt.show()
def main():
user_input = str(input("Please enter a text: "))
calculate_lists(user_input)
if __name__ == "__main__":
main()
I'm trying to implement a version of the huffman code in python. However, Im not able to add duplicate nodes to the graph. Is there a workaround to display values with the same text? To see what I mean, enter for example: aaaaabbbbcccdde
The graph only shows one node with the label 3.

Answer 1

I think you are mistaking nodes with node labels. Having duplicate nodes in a graph doesn't really make sense. What I feel you need here is to have duplicate labels.
What you can do to add the notion of labels to your graph is to have a dictionary that maps nodes identifiers (unique) to node labels (possibly not unique):
user_input = "aaaaabbbbcccdde"
# i is the node identifier and l is its corresponding label:
labels = {i: l for i, l in enumerate(user_input)}
nodes = labels.keys()
Using these you can construct your graph:
G = nx.DiGraph()
G.add_nodes_from(nodes)
Then you can, for example, draw it:
pos = nx.spring_layout(G)
nx.draw(G, pos)
nx.draw_networkx_labels(G, pos, labels)
And of course (probably most importantly), anytime you have a node identifier, say node_id, you can retrieve its label using labels[node_id]. What I suggest is to always work with node identifier, then at the very end, when you need to print a result you can translate node identifiers to something readable by a human, ie. node labels.
Depending on the complexity of your code, you may also find useful to attach the labels to the node objects themselves, networkx allows that:
nx.set_node_attributes(G, labels, 'label')
You'll then have access to node attributes:
for node_id, u in G.nodes(data=True):
print(u)
break
# Or if you have a node_identifier:
node_id = 1
print(G.node[node_id])
This would output:
{'label': 'a'}
{'label': 'a'}