P0-B: 数据损坏修复

- v1 上传大小校验: multipart body 长度与文件字节数不应相等,改为上限校验(上传功能恢复)
- mirror_sync 调度循环/remove_sync_source 死锁修复(锁外触发 start/stop)
- HTTP 断点续传损坏修复: 续传时先复制已有前缀再追加尾部,服务器忽略 Range 时从头重下
- sync_engine 同款续传 bug 同步修复;temp_path 提前初始化避免 except NameError

api/v1.py

@@ -1015,7 +1015,9 @@ class APIv1:
                                 total_written += len(chunk)
                                 file_size = total_written
                         
-                        if content_length > 0 and file_size != content_length:
+                        # content_length 是整个 multipart body 的长度（含 boundary/字段头），
+                        # 文件字节数不可能等于它；这里只做上限校验防止越界读取
+                        if content_length > 0 and file_size > content_length:
                             raise IOError(f"文件大小不匹配。期望: {content_length}, 实际: {file_size}")
                         
                         if os.path.exists(full_path):

core/mirror_sync.py

@@ -193,6 +193,7 @@ class MirrorSyncManager:
         while self.scheduler_running:
             try:
                 now = datetime.now()
+                due_sources = []
                 with self.sync_lock:
                     for name, status in self.sync_status.items():
                         source = self.sync_sources.get(name, {})
@@ -202,13 +203,21 @@ class MirrorSyncManager:
 
                         next_sync = status.get('next_sync')
                         if next_sync:
+                            try:
                                 next_time = datetime.fromisoformat(next_sync)
+                            except ValueError:
+                                continue
                             if now >= next_time:
-                                # 触发同步
+                                due_sources.append(name)
+
+                # 锁外触发同步，避免与 start_sync 内的 sync_lock 形成死锁
+                for name in due_sources:
+                    try:
                         print(f"[定时同步] 触发同步: {name}")
                         self.start_sync(name)
-                                # 计算下次同步时间
                         self._calculate_next_sync(name)
+                    except Exception as e:
+                        print(f"[定时同步] 触发 {name} 失败: {e}")
             except Exception as e:
                 print(f"定时同步调度错误: {e}")
             time.sleep(60)  # 每分钟检查一次
@@ -226,8 +235,12 @@ class MirrorSyncManager:
                 del self.sync_sources[name]
             if name in self.sync_status:
                 del self.sync_status[name]
+
+        # 锁外停止线程（stop_sync 内部会再取 sync_lock，避免不可重入锁死锁）
         if name in self.sync_threads:
             self.stop_sync(name)
+
+        with self.sync_lock:
             self.save_sync_state()
             self._save_sync_sources()
     
@@ -1118,11 +1131,14 @@ class MirrorSyncManager:
         return source_mtime > target_mtime or os.path.getsize(source_path) != os.path.getsize(target_path)
     
     def _download_file_http(self, url, local_path, config):
-        """下载HTTP文件"""
+        """下载HTTP文件（支持断点续传，续传数据追加到已有前缀，避免损坏）"""
         import urllib.request
         import base64
+        import shutil
         from email.utils import parsedate
 
+        temp_path = local_path + '.tmp'  # 提前初始化，避免 except 分支 NameError
+        
         try:
             req = urllib.request.Request(url)
             headers = config.get('headers', {})
@@ -1142,12 +1158,24 @@ class MirrorSyncManager:
                 if start_byte > 0:
                     req.add_header('Range', f'bytes={start_byte}-')
             
-            temp_path = local_path + '.tmp'
-            mode = 'ab' if start_byte > 0 else 'wb'
             
             with urllib.request.urlopen(req, timeout=timeout) as response:
-                if response.getcode() not in [200, 206]:
+                status_code = response.getcode()
-                    raise Exception(f"HTTP错误: {response.getcode()}")
+                if status_code not in [200, 206]:
+                    raise Exception(f"HTTP错误: {status_code}")
+
+                if start_byte > 0 and status_code != 206:
+                    # 服务器忽略了 Range，返回完整内容：从头开始写
+                    if os.path.exists(temp_path):
+                        os.remove(temp_path)
+                    start_byte = 0
+
+                mode = 'ab' if start_byte > 0 else 'wb'
+
+                if start_byte > 0:
+                    # 续传: 先把已有前缀复制进临时文件，再追加尾部数据，
+                    # 避免用"仅尾部"覆盖完整文件导致数据损坏
+                    shutil.copyfile(local_path, temp_path)
 
                 with open(temp_path, mode) as f:
                     while True:
@@ -1159,6 +1187,14 @@ class MirrorSyncManager:
                         f.write(chunk)
             
             if self.running:
+                # 校验续传结果: 续传后文件应不小于原有大小
+                if start_byte > 0:
+                    try:
+                        final_size = os.path.getsize(temp_path)
+                        if final_size < start_byte:
+                            raise Exception(f"续传后文件异常变小: {final_size} < {start_byte}")
+                    except OSError:
+                        raise
                 if os.path.exists(local_path):
                     os.remove(local_path)
                 os.rename(temp_path, local_path)

core/sync_engine.py

@@ -689,12 +689,25 @@ class SyncEngine:
                     req.add_header('Range', f'bytes={start_byte}-')
 
             timeout = config.get('timeout', 30)
-            temp_path = local_path + '.tmp'
+            temp_path = local_path + '.tmp'  # 提前初始化，避免 except 分支 NameError
-            mode = 'ab' if start_byte > 0 else 'wb'
 
             with urllib.request.urlopen(req, timeout=timeout) as response:
-                if response.getcode() not in [200, 206]:
+                status_code = response.getcode()
-                    raise Exception(f"HTTP错误: {response.getcode()}")
+                if status_code not in [200, 206]:
+                    raise Exception(f"HTTP错误: {status_code}")
+
+                if start_byte > 0 and status_code != 206:
+                    # 服务器忽略了 Range，返回完整内容：从头开始写
+                    if os.path.exists(temp_path):
+                        os.remove(temp_path)
+                    start_byte = 0
+
+                mode = 'ab' if start_byte > 0 else 'wb'
+
+                if start_byte > 0:
+                    # 续传: 先把已有前缀复制进临时文件，再追加尾部数据，避免损坏
+                    import shutil
+                    shutil.copyfile(local_path, temp_path)
 
                 total_size = int(response.headers.get('Content-Length', 0)) + start_byte
                 task.total_size = total_size