modify scripts

src/tools/sync_db.py

@@ -195,7 +195,7 @@ def generate_union_table(mysql_conn):
             key_str = parse_union_key(row["release_date"], row["studio_name"], row["whisper_code"])
             if key_str not in dict_whisper_data:
                 dict_whisper_data[key_str] = set()
-            dict_whisper_data[key_str].add(idx)
+            dict_whisper_data[key_str].add(row['whisper_code'].lower() if row['whisper_code'] else "")
 
         # 4. 内存中进行双边匹配
         result = []
@@ -210,6 +210,8 @@ def generate_union_table(mysql_conn):
             if w['release_date'] and w['studio_name']:
                 # 对name进行去空格，去特殊字符处理
                 key_str = parse_union_key(w['release_date'], w['studio_name'], w['whisper_code'])
+                whisper_codes = dict_whisper_data.get(key_str, set())
+                '''
                 if w['release_date'] == '2012-08-31' and w['studio_name'] == 'HEYZO':
                     print(f"whisper data: {w}")
                     print(f'key_str: {key_str}')
@@ -217,7 +219,7 @@ def generate_union_table(mysql_conn):
                         set_idx = dict_stash_data[key_str]
                         for idx in set_idx:
                             print(f"stash data: {stash_data[idx]}")
-
+                '''
                 if key_str in dict_stash_data:
                     set_idx = dict_stash_data[key_str]
                     for idx in set_idx:
@@ -229,6 +231,8 @@ def generate_union_table(mysql_conn):
                     if not matched:
                         # 否则匹配第一个未匹配过的
                         for idx in set_idx:
+                            if stash_data[idx]['code'].lower() in whisper_codes:
+                                continue
                             if not stash_data[idx].get('matched', False):
                                 s = stash_data[idx]
                                 stash_data[idx]['matched'] = True