fix(账号): 批量导入跳过已存在的重复账号

- 导入前按用户名（大小写不敏感）查库去重，同批文本内重复也跳过
- 返回消息区分格式跳过与重复跳过数量

web/backend/services/account_service.py

@@ -31,23 +31,36 @@ def cookie_account_ids_query(db: Session):
     ).distinct()
 
 
-def parse_and_build_accounts(text: str, default_tag: str = "") -> tuple[list[Account], int]:
+def parse_and_build_accounts(
+    db: Session,
+    text: str,
+    default_tag: str = "",
+) -> tuple[list[Account], int, int]:
     """
-    解析批量导入文本，构建 Account ORM 对象列表。
+    解析批量导入文本，构建 Account ORM 对象列表，并过滤已存在的重复账号。
 
     Args:
+        db: 数据库会话，用于查询已有账号
         text: 批量导入文本，每行格式：用户名|密码|邮箱|邮箱密码|标签(可选)
         default_tag: 统一标签兜底。某行未单独写标签时使用该值，
             行内第5列标签优先。
 
     Returns:
-        (accounts, skipped_count)
+        (accounts, skipped_count, duplicated_count)
+        skipped_count 为格式非法的行数；duplicated_count 为库中已存在或同批重复被跳过的行数。
     """
     from core.douyu.email_verifier import get_email_config_for_account
 
     default_tag = (default_tag or "").strip()
+    # 库中已有用户名（大小写不敏感），用于导入去重
+    existing_usernames = {
+        row[0].lower()
+        for row in db.query(Account.username).filter(Account.username != "").all()
+    }
     accounts = []
     skipped = 0
+    duplicated = 0
+    seen_in_batch: set[str] = set()
     for line in text.strip().split('\n'):
         line = line.strip()
         if not line or line.startswith('#'):
@@ -68,6 +81,12 @@ def parse_and_build_accounts(text: str, default_tag: str = "") -> tuple[list[Acc
             skipped += 1
             continue
 
+        username_key = username.lower()
+        if username_key in existing_usernames or username_key in seen_in_batch:
+            duplicated += 1
+            continue
+        seen_in_batch.add(username_key)
+
         email_cfg = get_email_config_for_account(email)
         accounts.append(Account(
             username=username,
@@ -79,4 +98,4 @@ def parse_and_build_accounts(text: str, default_tag: str = "") -> tuple[list[Acc
             email_imap_ssl=email_cfg.get('ssl', True),
             tag=tag,
         ))
-    return accounts, skipped
+    return accounts, skipped, duplicated