Updated axes. Added ignore-no-formulas logic.

src/components/plot-stats.py

@@ -8,8 +8,9 @@
 
 # identify "ExceLint-eligible" sheets!
 # there are at least two adjacent formulas (different or not)
-
-# number of suspicious ranges / number of rectangles
+#   number of suspicious ranges / number of rectangles
+# produce an ExceLint-eligible filter
+# also - omit any worksheets with no formulas (or "ineligible")
 
 parser = argparse.ArgumentParser('plot-stats.py')
 parser.add_argument('--input', help='Process an input .csv file, as generated by json-stats-to-csv.py.')
@@ -30,11 +31,14 @@
         workbookName = row['workbookName']
         worksheet = row['worksheet']
         suspiciousnessThreshold = row['suspiciousnessThreshold']
-        # Strip off .xlsx ending
-        items.append((workbookName[:-5] + '!' + worksheet, row['suspiciousCells']))
+        numFormulaCells = row['numFormulaCells']
+        # We don't bother with any sheet with one or fewer formulas, since ExceLint by definition will detect nothing.
+        if int(numFormulaCells) > 1:
+            # print(numFormulaCells)
+            # Strip off .xlsx ending
+            items.append((workbookName[:-5] + '!' + worksheet, row['suspiciousCells']))
 
 sorted_items = sorted(items, key=lambda x: int(x[1]))
-# print(sorted_items)
 
 for item in sorted_items:
     if int(item[1]) > 0:
@@ -45,7 +49,7 @@
 max_cells = cells[-1]
 
 plt.figure(figsize=(8,8))
-plt.axes([0.1,0.2,0.9,0.6])
+plt.axes([0.1,0.2,0.9,0.7])
 plt.title('ExceLint + CUSTODES: susp thresh = ' + str(suspiciousnessThreshold), y=1.02)
 #plt.xlabel('Worksheet name')
 plt.ylabel('# suspicious cells')