fix: skip already-processing UIDs at FETCH level to avoid downloading email content

main.py

@@ -48,7 +48,7 @@ def _email_poller(cfg):
     logger.info("邮件轮询线程已启动")
     while _running:
         try:
-            for acct_idx, mail in poll_accounts(cfg):
+            for acct_idx, mail in poll_accounts(cfg, skip_uids=_processing_uids):
                 if not _running:
                     return
                 with _pending_lock:

src/email_client.py

@@ -109,8 +109,10 @@ def _login_and_prepare(account: EmailAccount, timeout: int = 30, select_inbox: b
     return conn
 
 
-def fetch_unseen_emails(account: EmailAccount) -> list[Email]:
+def fetch_unseen_emails(account: EmailAccount, skip_uids: set[bytes] | None = None) -> list[Email]:
     conn = _login_and_prepare(account, select_inbox=False)
+    if skip_uids is None:
+        skip_uids = set()
 
     # 列出所有文件夹
     _, folder_data = conn.list()
@@ -120,8 +122,6 @@ def fetch_unseen_emails(account: EmailAccount) -> list[Email]:
             if item is None:
                 continue
             line = item.decode("utf-8", errors="replace")
-            # IMAP LIST 格式: (flags) "delimiter" "folder_name"
-            # 用正则提取最后一个引号内的内容
             m = re.search(r'"([^"]*)"\s*$', line)
             if m:
                 folders.append(m.group(1))
@@ -130,6 +130,7 @@ def fetch_unseen_emails(account: EmailAccount) -> list[Email]:
     logger.info("扫描文件夹: %s", ", ".join(folders))
 
     emails = []
+    skipped = 0
     for folder in folders:
         try:
             status, _ = conn.select(folder, readonly=True)
@@ -140,9 +141,15 @@ def fetch_unseen_emails(account: EmailAccount) -> list[Email]:
             uids = data[0].split() if data[0] else []
             if not uids:
                 continue
-            logger.info("  %s: %d 封未读", folder, len(uids))
 
-            for uid in uids:
+            # 先过滤掉正在处理的 UID，避免下载不需要的邮件
+            new_uids = [uid for uid in uids if uid not in skip_uids]
+            skipped += len(uids) - len(new_uids)
+            if not new_uids:
+                continue
+            logger.info("  %s: %d 封未读 (%d 封跳过)", folder, len(new_uids), len(uids) - len(new_uids))
+
+            for uid in new_uids:
                 try:
                     _, msg_data = conn.uid("FETCH", uid, "RFC822")
                     if msg_data[0] is None:

src/summarizer.py

@@ -8,11 +8,11 @@ from src.tg_bot import send_summary, send_thinking, edit_message, format_summary
 logger = logging.getLogger(__name__)
 
 
-def poll_accounts(cfg: Config) -> Generator[tuple[int, Email], None, None]:
+def poll_accounts(cfg: Config, skip_uids: set[bytes] | None = None) -> Generator[tuple[int, Email], None, None]:
     for idx, acct in enumerate(cfg.email_accounts):
         try:
             logger.info(f"检查邮箱: {acct.username}")
-            emails = fetch_unseen_emails(acct)
+            emails = fetch_unseen_emails(acct, skip_uids=skip_uids)
             if emails:
                 logger.info(f"  发现 {len(emails)} 封新邮件")
             for mail in emails:

src/tg_bot.py

@@ -193,7 +193,8 @@ def _summary_keyboard(can_reply: bool = True, is_promotion: bool = False, links:
     for link in (links or []):
         url = link.get("url", "")
         text = link.get("text", "打开链接")[:30]
-        if url.startswith("http"):
+        # 校验 URL：合法 http/https，无多重协议头，长度合理
+        if url.startswith("http") and url.count("://") == 1 and len(url) < 200:
             kb.append([{"text": f"🔗 {text}", "url": url}])
     return {"inline_keyboard": kb}