fix: 浏览内容/附件进度实时更新

api_browser.py

@@ -230,7 +230,7 @@ class APIBrowser:
     def get_article_list_page(self, bz: int = 2, page: int = 1, base_url: str = None):
         """获取单页文章列表"""
         if not self.logged_in:
-            return [], 0, None
+            return [], 0, None, 0
 
         try:
             if base_url and page > 1:
@@ -264,9 +264,10 @@ class APIBrowser:
                             'article_id': article_id,
                         })
 
-            # 获取总页数
+            # 获取总页数 / 总记录数
             total_pages = 1
             next_page_url = None
+            total_records = 0
 
             page_content = soup.find(id='PageContent')
             if page_content:
@@ -282,11 +283,11 @@ class APIBrowser:
                     if next_href:
                         next_page_url = f"{BASE_URL}/admin/{next_href}"
 
-            return articles, total_pages, next_page_url
+            return articles, total_pages, next_page_url, total_records
 
         except Exception as e:
             self.log(f"[API] 获取列表失败: {str(e)}")
-            return [], 0, None
+            return [], 0, None, 0
 
     def get_article_attachments(self, article_href: str):
         """获取文章的附件列表"""
@@ -338,14 +339,19 @@ class APIBrowser:
         except:
             return False
 
-    def browse_content(self, browse_type: str,
+    def browse_content(
-                       should_stop_callback: Optional[Callable] = None) -> APIBrowseResult:
+        self,
+        browse_type: str,
+        should_stop_callback: Optional[Callable] = None,
+        progress_callback: Optional[Callable] = None,
+    ) -> APIBrowseResult:
         """
         浏览内容并标记已读
 
         Args:
             browse_type: 浏览类型 (应读/注册前未读)
             should_stop_callback: 检查是否应该停止的回调函数
+            progress_callback: 进度回调（可选），用于实时上报已处理/总数
 
         Returns:
             浏览结果
@@ -368,13 +374,14 @@ class APIBrowser:
         self.log(f"[API] 开始浏览 '{browse_type}' (bz={bz})...")
 
         try:
-            total_items = 0
+            browsed_items = 0
-            total_attachments = 0
+            viewed_attachments = 0
+            discovered_attachments = 0
             page = 1
             base_url = None
 
             # 获取第一页
-            articles, total_pages, next_url = self.get_article_list_page(bz, page)
+            articles, total_pages, next_url, total_records = self.get_article_list_page(bz, page)
 
             if not articles:
                 self.log(f"[API] '{browse_type}' 没有待处理内容")
@@ -386,6 +393,32 @@ class APIBrowser:
             if next_url:
                 base_url = next_url
 
+            last_report_ts = 0.0
+
+            def report_progress(force: bool = False):
+                nonlocal last_report_ts
+                if not progress_callback:
+                    return
+                now_ts = time.time()
+                if not force and now_ts - last_report_ts < 1.0:
+                    return
+                last_report_ts = now_ts
+                try:
+                    progress_callback(
+                        {
+                            "total_items": int(total_records or 0),
+                            "browsed_items": int(browsed_items or 0),
+                            "total_attachments": int(discovered_attachments or 0),
+                            "viewed_attachments": int(viewed_attachments or 0),
+                            "page": int(page or 0),
+                            "total_pages": int(total_pages or 0),
+                        }
+                    )
+                except Exception:
+                    pass
+
+            report_progress(force=True)
+
             # 处理所有页面
             while True:
                 if should_stop_callback and should_stop_callback():
@@ -397,17 +430,20 @@ class APIBrowser:
                         break
 
                     title = article['title'][:30]
-                    total_items += 1
+                    browsed_items += 1
 
                     # 获取附件
                     attachments = self.get_article_attachments(article['href'])
 
                     if attachments:
+                        discovered_attachments += len(attachments)
                         for attach in attachments:
                             if self.mark_read(attach['id'], attach['channel_id']):
-                                total_attachments += 1
+                                viewed_attachments += 1
 
-                        self.log(f"[API] [{total_items}] {title} - {len(attachments)}个附件")
+                        self.log(f"[API] [{browsed_items}] {title} - {len(attachments)}个附件")
+
+                    report_progress()
 
                     time.sleep(0.1)
 
@@ -416,20 +452,22 @@ class APIBrowser:
                 if page > total_pages:
                     break
 
-                articles, _, next_url = self.get_article_list_page(bz, page, base_url)
+                articles, _, next_url, _ = self.get_article_list_page(bz, page, base_url)
                 if not articles:
                     break
 
                 if next_url:
                     base_url = next_url
 
+                report_progress(force=True)
                 time.sleep(0.2)
 
-            self.log(f"[API] 浏览完成: {total_items} 条内容，{total_attachments} 个附件")
+            report_progress(force=True)
+            self.log(f"[API] 浏览完成: {browsed_items} 条内容，{viewed_attachments} 个附件")
 
             result.success = True
-            result.total_items = total_items
+            result.total_items = browsed_items
-            result.total_attachments = total_attachments
+            result.total_attachments = viewed_attachments
             return result
 
         except Exception as e:

services/tasks.py

@@ -590,12 +590,46 @@ def run_task(user_id, account_id, browse_type, enable_screenshot=True, source="m
 
                             safe_update_task_status(account_id, {"detail_status": "正在浏览"})
                             log_to_client(f"开始浏览 '{browse_type}' 内容...", user_id, account_id)
+                            account.total_items = 0
+                            account.total_attachments = 0
+                            safe_update_task_status(account_id, {"progress": {"items": 0, "attachments": 0}})
 
                             def should_stop():
                                 return account.should_stop
 
+                            last_progress_ts = 0.0
+
+                            def on_browse_progress(progress: dict):
+                                nonlocal last_progress_ts
+                                try:
+                                    now_ts = time_module.time()
+                                    if now_ts - last_progress_ts < 0.5:
+                                        return
+                                    last_progress_ts = now_ts
+
+                                    total_items = int(progress.get("total_items") or 0)
+                                    browsed_items = int(progress.get("browsed_items") or 0)
+                                    total_attachments = int(progress.get("total_attachments") or 0)
+                                    viewed_attachments = int(progress.get("viewed_attachments") or 0)
+
+                                    if total_items > 0:
+                                        account.total_items = total_items
+                                    else:
+                                        account.total_items = max(int(getattr(account, "total_items", 0) or 0), browsed_items)
+
+                                    account.total_attachments = max(total_attachments, viewed_attachments)
+                                    safe_update_task_status(
+                                        account_id, {"progress": {"items": browsed_items, "attachments": viewed_attachments}}
+                                    )
+                                except Exception:
+                                    pass
+
                             checkpoint_mgr.update_stage(task_id, TaskStage.BROWSING, progress_percent=50)
-                            result = api_browser.browse_content(browse_type=browse_type, should_stop_callback=should_stop)
+                            result = api_browser.browse_content(
+                                browse_type=browse_type,
+                                should_stop_callback=should_stop,
+                                progress_callback=on_browse_progress,
+                            )
                         else:
                             error_message = "登录失败"
                             log_to_client(f"❌ {error_message}", user_id, account_id)