Added limit to minimum frequency = 10 + Ordered by frequency

scripts/recalculate_statistics.py

@@ -174,6 +174,12 @@ def main(args):
         write_file_path = os.path.join(args.output, file_name)
         with open(read_file_path, 'r') as rf, open(write_file_path, 'w') as wf:
             original_text, stats = get_new_stats(rf)
+            freq_pos = original_text[0].index('Frequency')
+            original_text = [original_text[0]] + [l for l in original_text[1:] if int(l[freq_pos]) >= 10]
+            if len(original_text) > 1:
+                original_text = [original_text[0]] + sorted(original_text[1:], key=lambda x: -1 * int(x[freq_pos]))
+            else:
+                original_text = [original_text[0]]
             write_new_stats(wf, original_text, stats, file_name, word_order)
 
 if __name__ == '__main__':