feat: 增强错误处理和日志记录，优雅处理通道关闭情况

src/bootstrap.rs

@@ -1,3 +1,26 @@
 pub fn initialize_process_runtime() {
     let _ = rustls::crypto::ring::default_provider().install_default();
+
+    // Install a global panic hook so that any panic in a spawned task
+    // is logged with a full backtrace before the process exits.  Without
+    // this hook, panics are silent (or print a brief message to stderr)
+    // which makes root-causing crashes difficult.
+    let default_hook = std::panic::take_hook();
+    std::panic::set_hook(Box::new(move |info| {
+        // Use tracing so the panic appears in the same log stream as everything else.
+        tracing::error!(
+            panic.payload = ?info.payload(),
+            panic.location = ?info.location(),
+            "FATAL: process panicked — collecting backtrace"
+        );
+        // Print a compact backtrace to stderr as well (backtrace is not
+        // captured by tracing).
+        let backtrace = std::backtrace::Backtrace::capture();
+        if backtrace.status() == std::backtrace::BacktraceStatus::Captured {
+            eprintln!("FATAL panic backtrace:\n{}", backtrace);
+        }
+        // Delegate to the default hook which prints the panic message and
+        // optionally the RUST_BACKTRACE-based backtrace.
+        default_hook(info);
+    }));
 }

src/bus/mod.rs

@@ -43,18 +43,13 @@ impl MessageBus {
             .map_err(|_| BusError::Closed)
     }
 
-    /// Consume a message from the inbound queue
+    /// Consume a message from the inbound queue.
-    pub async fn consume_inbound(&self) -> InboundMessage {
+    /// Returns `None` when the channel is closed (all senders dropped).
-        let msg = self
+    pub async fn consume_inbound(&self) -> Option<InboundMessage> {
-            .inbound_rx
+        let msg = self.inbound_rx.lock().await.recv().await?;
-            .lock()
-            .await
-            .recv()
-            .await
-            .expect("bus inbound closed");
         #[cfg(debug_assertions)]
         tracing::debug!(channel = %msg.channel, sender = %msg.sender_id, chat = %msg.chat_id, "Bus: consuming inbound message");
-        msg
+        Some(msg)
     }
 
     /// Publish a message to the outbound queue
@@ -67,14 +62,10 @@ impl MessageBus {
             .map_err(|_| BusError::Closed)
     }
 
-    /// Consume an outbound message from the outbound queue
+    /// Consume an outbound message from the outbound queue.
-    pub async fn consume_outbound(&self) -> OutboundMessage {
+    /// Returns `None` when the channel is closed (all senders dropped).
-        self.outbound_rx
+    pub async fn consume_outbound(&self) -> Option<OutboundMessage> {
-            .lock()
+        self.outbound_rx.lock().await.recv().await
-            .await
-            .recv()
-            .await
-            .expect("bus outbound closed")
     }
 }
 

src/channels/wechat.rs

@@ -8,6 +8,7 @@ use std::sync::{
 use std::time::UNIX_EPOCH;
 
 use async_trait::async_trait;
+use futures_util::FutureExt;
 use tokio::sync::RwLock;
 use tokio::task::JoinHandle;
 use wechatbot::{BotOptions, SendContent, WeChatBot};
@@ -246,6 +247,12 @@ impl Channel for WechatChannel {
         let running = self.running.clone();
 
         let handle = tokio::spawn(async move {
+            // Use catch_unwind to prevent a panic in the WeChat SDK (login or
+            // long-poll loop) from crashing the entire process.  Any panic is
+            // logged and the channel is cleanly marked as stopped.
+            // AssertUnwindSafe is needed because WeChatBot contains internal
+            // locks (RwLock) that are not RefUnwindSafe.
+            let result = std::panic::AssertUnwindSafe(async {
                 match bot.login(force_login).await {
                     Ok(creds) => {
                         tracing::info!(
@@ -256,7 +263,6 @@ impl Channel for WechatChannel {
                         );
                     }
                     Err(error) => {
-                    running.store(false, Ordering::SeqCst);
                         tracing::error!(channel = %channel_name, error = %error, "WeChat login failed");
                         return;
                     }
@@ -265,6 +271,16 @@ impl Channel for WechatChannel {
                 if let Err(error) = bot.run().await {
                     tracing::error!(channel = %channel_name, error = %error, "WeChat channel stopped with error");
                 }
+            })
+            .catch_unwind()
+            .await;
+
+            if let Err(_panic) = result {
+                tracing::error!(
+                    channel = %channel_name,
+                    "WeChat bot task panicked — marking channel as stopped"
+                );
+            }
 
             running.store(false, Ordering::SeqCst);
         });

src/gateway/outbound_dispatcher.rs

@@ -34,7 +34,13 @@ impl OutboundDispatcher {
         tracing::info!("OutboundDispatcher started");
 
         loop {
-            let msg = self.bus.consume_outbound().await;
+            let msg = match self.bus.consume_outbound().await {
+                Some(msg) => msg,
+                None => {
+                    tracing::info!("Outbound bus closed, stopping dispatcher");
+                    break;
+                }
+            };
             #[cfg(debug_assertions)]
             tracing::debug!(
                 channel = %msg.channel,

src/gateway/processor.rs

@@ -114,8 +114,14 @@ impl InboundProcessor {
         );
 
         loop {
-            // 1. 消费消息
+            // 1. 消费消息 (channel 关闭时返回 None，优雅退出)
-            let inbound = self.bus.consume_inbound().await;
+            let inbound = match self.bus.consume_inbound().await {
+                Some(msg) => msg,
+                None => {
+                    tracing::info!("Inbound bus closed, stopping inbound processor");
+                    break;
+                }
+            };
 
             #[cfg(debug_assertions)]
             {

src/gateway/session.rs

@@ -1753,7 +1753,8 @@ mod tests {
 
         let msg = tokio::time::timeout(std::time::Duration::from_millis(500), bus.consume_outbound())
             .await
-            .expect("should have received an outbound message");
+            .expect("timeout waiting for outbound message")
+            .expect("bus outbound closed");
         assert_eq!(msg.event_kind, OutboundEventKind::ToolResult);
     }
 

src/gateway/session_message_sender.rs

@@ -138,7 +138,7 @@ mod tests {
             }
         );
 
-        let msg = bus.consume_outbound().await;
+        let msg = bus.consume_outbound().await.expect("bus outbound closed");
         assert_eq!(msg.content, "hello");
         assert_eq!(msg.media.len(), 1);
         assert_eq!(msg.media[0].media_type, "image");

src/scheduler/mod.rs

@@ -1681,7 +1681,8 @@ mod tests {
             bus.consume_outbound(),
         )
         .await
-        .unwrap();
+        .expect("timeout waiting for outbound message")
+        .expect("bus outbound closed");
         assert_eq!(outbound.channel, "test-channel");
         assert_eq!(outbound.chat_id, "oc_demo");
         assert!(outbound.content.contains("定时任务执行失败"));