fixed some sturf

Python/collectvariables.py

@@ -18,12 +18,15 @@ def get_unique_vars(filename):
 def main():
 	count = 0
 	for filename in glob.glob("pydata/pyfile*.txt"):
-		print("working on " + filename)
-		s = get_unique_vars(filename)
-		with open("pyvariables/varfile" + str(count) + ".txt", "w") as f:
-			for var in s:
-				f.write(var + "\n")
-		count += 1
+		if count == 50000:
+			return
+		else:
+			print("working on " + filename)
+			s = get_unique_vars(filename)
+			with open("pyvariables/varfile" + str(count) + ".txt", "w") as f:
+				for var in s:
+					f.write(var + "\n")
+			count += 1
 
 
 if __name__ == '__main__':

Python/toptext.py

@@ -17,7 +17,7 @@ class MRWordFreqCount(MRJob):
     def mapper(self, _, line):
         file1, file2, total = line.split(",")
         file2name = "pyfile" + file2 + ".txt"
-        for file_number in {file1, file2}:
+        for file_number in {file1, file2}: #If file1 == file2
         	filename = "pyfile" + file_number + ".txt"
         	file_text = filename.read()
         	for i in range(int(file1) + 1, int(total)):