Code cleanup.

scansize.py

@@ -13,19 +13,16 @@ import pandas as pd
 import numpy as np
 import matplotlib.pyplot as plt
 
-
 debug = ''
 
-runsizes = dict()
-
-def parse_bnx(file_path, size_dict = {}):
+def parse_bnx(runsizes = {}):
     for line in fileinput.input():
         if line.startswith('# Run '):
             if debug : print line.strip()
             line = line.split('\t')
             # The SourceFolder of this Run is the 2nd entry of the line
-            if debug : SourceFolder = line[1]
-            print SourceFolder
+            SourceFolder = line[1]
+            if debug : print SourceFolder
             # The RunId of this Run is the last entry of the line
             RunId = int(line[-1])
             if debug : print RunId
@@ -34,9 +31,7 @@ def parse_bnx(file_path, size_dict = {}):
             NumberofScans = int(line[7])
             for Scan in range(1, NumberofScans + 1):
                 runsizes[RunId][Scan] = 0
-                size_dict[Scan] = 0
             if debug : print runsizes
-            if debug : print size_dict
         elif line.startswith('0\t'):
             line = line.split('\t')
             if debug : print line
@@ -48,24 +43,13 @@ def parse_bnx(file_path, size_dict = {}):
             RunId = int(line[-2])
             if debug: print ScanNumber, Length
             runsizes[RunId][ScanNumber] += Length
-            size_dict[ScanNumber] += Length
             if debug : print runsizes
-            if debug : print size_dict
-    return size_dict
-
-def hist_plot(size_dict, values = []):
-    for n in size_dict:
-        values.append(size_dict[n])
-    plt.plot(values)
-    plt.show()
+    return runsizes
 
 if __name__ == '__main__':
-    parse_bnx(sys.argv[1])
-    # print size_dict
+    runsizes = dict()
+    runsizes = parse_bnx()
     print json.dumps(runsizes,sort_keys=True,indent=4, separators=(',', ': '))
-    labels = runsizes.keys() # 1,2,..
-
-    # hist_plot(size_dict)
 
 """