run_pagerank.py

import networkx as nx

import soundcloud
from sc_pagerank import computePR
import sc_api_calls as scac

artistGraph = nx.MultiDiGraph()

try:
    print("Reading in artist graph...")
    artistGraph = nx.read_pajek("artistGraph.net")
    print("Read successfully!")
    print("The artist graph currently contains " + str(len(artistGraph)) + " artists.")
    print(
        "The artist graph currently contains "
        + str(nx.number_strongly_connected_components(artistGraph))
        + " strongly connected components."
    )
except IOError:
    print("Could not find artistGraph")

client = soundcloud.Client(client_id="454aeaee30d3533d6d8f448556b50f23")

raw_name = input("Enter a soundcloud artist to analyze: ")


# Artist of interest
search = client.get("/users/", q=raw_name)[0]

print("Artist interpreted as: %s" % search.username)
# need to compute all neighbors in given graph selection before we can compute the
# pr of each node.
print("=" * 20)

my_component = artistGraph

for component in nx.strongly_connected_component_subgraphs(artistGraph):
    if search.id in component:
        my_component = component

# Go through the graph and compute each PR until it converges.
iterations = 10
print("Computing PageRank on your searched artist...")
computePR(my_component, 0.85, iterations)

prList = []

for artist in my_component.nodes():
    prList.append((artist, my_component.node[artist]["currPR"]))

prList.sort(key=lambda tup: tup[1])  # Sort the list in place

prList.reverse()  # order by descending PR

print(("Here are some artists similar to " + str(search.username)))

for item in prList[0:10]:
    artist = scac.id2username(item[0])
    print(artist, item[1])