fixed merge

Python/pyfilesplit.py

 import re
+import sys
 
 def go(subset):
 	'''

Python/pyfuncsplit.py

 import re
+import sys
 
 def go(subset):
 	'''

Python/topfunctions.py

@@ -4,7 +4,7 @@ import heapq
 import jellyfish
 
 #Sek k of top-k entries to collect.
-CAPACITY = 200
+CAPACITY = 500
 
 class TopFunctions(MRJob):
     def mapper(self, _, line):

Python/topvariables.py

@@ -6,8 +6,6 @@ import math
 
 CAPACITY = 500
 
-VARIABLE_REGEX = ("[\s]*([\w, ]+)[\s]*=")
-
 class TopVariables(MRJob):
 
     def mapper(self, _, line):
@@ -91,7 +89,7 @@ class TopVariables(MRJob):
         for i in range(len(self.h)):
             item = heapq.heappop(self.h)
             #Yield the mean scores.
-            yield item[1], -item[0] / self.total
+            yield item[1], -item[0] / (self.total - 1)
 
 if __name__ == '__main__':
     TopVariables.run()
\ No newline at end of file

text_to_dict.py

@@ -17,7 +17,7 @@ def to_json(filename, dataname):
             #Converts each line into a dictonary entry.
             name, score = result.split("\t")
             name = name.replace('"', '')
-            score = score.strip()
-            resultdict[name] = float(score)
+            score = score.strip() 
+            resultdict[name] = float(score) * 100
     with open(dataname, 'w') as datafile:
         json.dump(resultdict, datafile)
\ No newline at end of file