sort the dict, update function name

Former-commit-id: 5b410d8b2bc51ac057516c7a81c273849a60d6b2

dict_util.py

@@ -4,33 +4,39 @@ import pickle
 ### NOTICE: csv only accept two colomn input. but accept multi-time input.
 
 
-# 1_2_3, 1 is action, 2 is supply object, 3 is accept object
-def update_dict_csv(term_dict, f):
+# 1_2_3, 1 is action, 2 is supply object, 3 is source object
+def update_dict_csv(term_dict:dict, f):
     for rows in csv.reader(f):
         word = rows[0].lower()
         if word in term_dict:
             if rows[1] not in term_dict[word]:
                 term_dict[word] = term_dict[word]+[rows[1]]
+            else:
+                print("{},{}已存在".format(word, rows[1]))
         else:
             term_dict[word]=[rows[1]]
+    term_dict = sort_dict(term_dict)
     pass
 
-def export_dict_csv(term_dict, f):
+def export_csv_dict(term_dict:dict, f):
     for key, val in term_dict.items():
         csv.writer(f).writerow([key, val])
     pass
 
-def save_dict_pickle(term_dict, f):
+def save_pickle_dict(term_dict:dict, f):
     pickle.dump(term_dict, f, pickle.HIGHEST_PROTOCOL)
     pass
 
-def update_csv_pickle(pickle_f, csv_f):
+def update_pickel_csv(pickle_f, csv_f):
     term_dict = pickle.load(pickle_f)
     update_dict_csv(term_dict, csv_f)
-    #save to pickle file, highest protocal to get better performance
+    #save to pickle file, highest protocol to get better performance
     pickle.dump(term_dict, pickle_f, pickle.HIGHEST_PROTOCOL)
     pass
 
+def sort_dict(term_dict:dict):
+    term_dict = dict(sorted(term_dict.items(), key=lambda x:len(x[0]), reverse=True))
+    return term_dict
 
 #demo
 term_dict_sc2 = {}
@@ -38,7 +44,7 @@ with open("./finetune_data/dict_enzh.csv", 'r', encoding='utf-8') as f:
     update_dict_csv(term_dict_sc2,f)
 
 with open("../test.csv", "w", encoding='utf-8') as w:
-    export_dict_csv(term_dict_sc2,w)
+    export_csv_dict(term_dict_sc2,w)
 
 ## for load pickle, just:
 # pickle.load(f)