fixed directed pagerank calculations

pageRank.py

@@ -22,9 +22,9 @@ def rank(self):
             if self.directed:
                 neighbors = self.graph.out_edges(key)
                 for n in neighbors:
-                    outlinks = len(self.graph.out_edges(n[0]))
-                    rank_sum += (1 / float(outlinks)) * self.ranks[n[0]]
-
+                    outlinks = len(self.graph.out_edges(n[1]))
+                    if outlinks > 0:
+                        rank_sum += (1 / float(outlinks)) * self.ranks[n[1]]
             else: 
                 neighbors = self.graph[key]
                 for n in neighbors:
@@ -49,7 +49,6 @@ def rank(self):
         graph = parse(filename, isDirected)
         p = PageRank(graph, isDirected)
         p.rank()
-
         for state, rank in p.ranks.iteritems():
             print state+": "+str(rank)
 

utils.py

@@ -1,4 +1,4 @@
-import sys, math, random, csv, types, networkx as nx
+import re, sys, math, random, csv, types, networkx as nx
 from collections import defaultdict
 
 def parse(filename, isDirected):
@@ -27,10 +27,11 @@ def parse_directed(data):
     DG = nx.DiGraph()
 
     for i, row in enumerate(data):
-        node_a = row[0].strip()
-        node_b = row[2].strip()
-        val_a = int(row[1])
-        val_b = int(row[3])
+
+        node_a = format_key(row[0])
+        node_b = format_key(row[2])
+        val_a = digits(row[1])
+        val_b = digits(row[3])
 
         DG.add_edge(node_a, node_b)
         if val_a >= val_b:
@@ -40,6 +41,16 @@ def parse_directed(data):
 
     return DG
 
+def digits(val):
+    return int(re.sub("\D", "", val))
+
+def format_key(key):
+    key = key.strip() 
+    if key.startswith('"') and key.endswith('"'):
+        key = key[1:-1]
+    return key 
+
+
 def print_results(f, method, results):
     print method