'''create a graph with all the edges that are not in the original graph

iterate the list of the users and for every user adds all the edges that are not already in

the original graph.

Args:

g: The original graph as we created it according to the given csv

users: Dictionary of the users.

Returns:

f_f_edge_list: The Complement graph of g

'''

if not g :

res = open('other_edges.dump', 'r')

f_edge_list = pickle.load(res)

res.close()

else:

f_edge_list = nx.Graph()

i=0

for u in users:

for f in users:

if (not f == u) and (not (f in g.neighbors(u))):

f_edge_list.add_edge(u, f)

i+=1

r = str(i) + "\r"

stdout.write(r)

res_f = open('other_edges.dump', 'w')

pickle.dump(f_edge_list,res_f )

res_f.close()

'''create graph with all the edges that are in the list edges (which is constructed)

from the csv file we received as input.

Args:

edges: list of lists in length 2 [source, target]

Returns:

G: a networkx graph data structure

f_edge_list: The Complement graph of G

'''

if not edges:

reso = open('initial_graph.dump', 'r')

G = pickle.load(reso)

reso.close()

f_edge_list = make_future_edges()

else:

with open('user_artists.csv', 'rb') as csvfile:

all_artist_data = list(csv.reader(csvfile))[1:]

G = nx.Graph()

users = {} # Dictionary where user IDs are kept as keys

for edge in edges:

users[edge[0]] = True

nodes_to_add = []

for u in all_artist_data:

if not u[0] in users:

users[u[0]] = True

nodes_to_add.append(u[0])

for (s, t) in edges:

G.add_edge(s, t)

G.add_nodes_from(nodes_to_add)

res_f = open('initial_graph.dump', 'w')

pickle.dump(G,res_f )

res_f.close()

f_edge_list = make_future_edges(G,users)

# artist_IDs = ['70', '150', '989', '16326', '144882', '194647', '389445', '390392', '511147', '532992'] #List of artists to choose from

# artist_IDs = ['150', '989', '194647', '390392', '511147', '532992'] #List of artists to choose from

"""Class that reads the data from the files user_friends.csv and

user_artists.csv into self.edges and self.user_artist_data,

respectively. self.edges is a list of lists of size two that

represent edges. user_artists.csv is a list of lists of size three

[userID, artistID, times_was_heared]."""

def __init__(self, my_artists=['150', '989', '194647', '390392', '511147', '532992']):

self.artist_IDs = my_artists

with open('user_friends.csv', 'rb') as csvfile:

self.edges = list(csv.reader(csvfile))[1:]

print "edges done"

with open('user_artists.csv', 'rb') as csvfile:

all_artist_data = list(csv.reader(csvfile))[1:]

self.user_artist_data = [x for x in all_artist_data if x[1] in self.artist_IDs]

print "user_artist_data done"

Gg, Cg = create_graph(self.edges)

self.graph = Gg

print "graph done"

self.c_graph = Cg

print "c_graph done"

t_users = {}

for user_n in Gg.nodes():

name = user_n

artists = self.get_artists(name) #TODO: implement

t_users[name] = user.user(name, artists)

self.users = t_users

print "users done"

self.cds_sold = 0

self.art_analisys = {}

for aa in self.artist_IDs:

self.art_analisys[aa] = 0

def get_artists(self, name):

'''calculate how many time the users listened to the artist

:param name: name of the user

:return: dictionary {'artist_ID' : number of time listened}

'''

art_dic = {}

for art_name in self.artist_IDs:

art_dic[art_name] = 0

for u ,a,l in self.user_artist_data:

if u == str(name):

art_dic[a] += int(l)

# print name, a, l

return art_dic

def update_edges(self):

'''

this function iterarte over the complement graph and check for every edge

if we need to add it or not according to the probability and how many common friends.

change the actual graph (Network.graph)

:return: the number of edges that we added

'''

P = 0.02

edges_to_add = []

for e in self.c_graph.edges():

l1 = set(self.graph.neighbors(e[0])) #get all neighbors of s

l2 = set(self.graph.neighbors(e[1])) #get all neighbors of t

lc = l1.intersection(l2)

k = len(lc)

prob = 1-pow((1-P), k)

r = random()

if r < prob:

edges_to_add.append(e)

self.graph.add_edges_from(edges_to_add)

self.c_graph.remove_edges_from(edges_to_add)

return len(edges_to_add)

def move_time(self, n=1):

'''

clculate a time lap of the system calculate the new friendships and the CDs that users buy.

the function only update the network at the end of the epoch.

:param n: number of epochs to run

:return: tuple of (num of CDs, num of edges, list of updates for the network.)

'''

for i in range(n):

updates = []

j = 0

for u in self.users:

j += 1

r = str(j) + " \r"

stdout.write(r)

cur_friends = self.graph.neighbors(u)

new_disks_names = self.users[u].calc_new_disk(cur_friends,self.users, self.artist_IDs)

if len(new_disks_names) > 0:

updates.append((u,new_disks_names))

for d in new_disks_names:

self.art_analisys[d] += 1

e_added = self.update_edges()

# print e_added, "edges was added"

tmp_sold=0

for us,ds in updates:

tmp_sold += len(ds)

self.users[us].add_disks(ds)

# print "user {} bought the following CDs".format(us), ds

self.cds_sold += tmp_sold