fix(desktop): DeerFlow UI — ChatArea refactor + ai-elements + dead CSS cleanup

ChatArea retry button uses setInput instead of direct sendToGateway,
fix bootstrap spinner stuck for non-logged-in users,
remove dead CSS (aurora-title/sidebar-open/quick-action-chips),
add ai components (ReasoningBlock/StreamingText/ChatMode/ModelSelector/TaskProgress),
add ClassroomPlayer + ResizableChatLayout + artifact panel

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

crates/zclaw-saas/src/workers/generate_embedding.rs

@@ -44,13 +44,7 @@ impl Worker for GenerateEmbeddingWorker {
             }
         };
 
-        // 2. 删除旧分块（full refresh on each update）
-        sqlx::query("DELETE FROM knowledge_chunks WHERE item_id = $1")
-            .bind(&args.item_id)
-            .execute(db)
-            .await?;
-
-        // 3. 分块
+        // 2. 分块
         let chunks = crate::knowledge::service::chunk_content(&content, 512, 64);
 
         if chunks.is_empty() {
@@ -58,13 +52,32 @@ impl Worker for GenerateEmbeddingWorker {
             return Ok(());
         }
 
-        // 4. 写入分块（带关键词继承）
+        // 3. 在事务中删除旧分块 + 插入新分块（防止并发竞争条件）
+        let mut tx = db.begin().await?;
+
+        // 锁定条目行防止并发 worker 同时处理同一条目
+        let locked: Option<(String,)> = sqlx::query_as(
+            "SELECT id FROM knowledge_items WHERE id = $1 FOR UPDATE"
+        )
+        .bind(&args.item_id)
+        .fetch_optional(&mut *tx)
+        .await?;
+
+        if locked.is_none() {
+            tx.rollback().await?;
+            tracing::warn!("GenerateEmbedding: item {} was deleted during processing", args.item_id);
+            return Ok(());
+        }
+
+        sqlx::query("DELETE FROM knowledge_chunks WHERE item_id = $1")
+            .bind(&args.item_id)
+            .execute(&mut *tx)
+            .await?;
+
         for (idx, chunk) in chunks.iter().enumerate() {
             let chunk_id = uuid::Uuid::new_v4().to_string();
 
-            // 为每个 chunk 提取额外关键词（简单策略：标题 + 继承关键词）
             let mut chunk_keywords = keywords.clone();
-            // 从 chunk 内容提取高频词作为补充关键词
             extract_chunk_keywords(chunk, &mut chunk_keywords);
 
             sqlx::query(
@@ -76,10 +89,12 @@ impl Worker for GenerateEmbeddingWorker {
             .bind(idx as i32)
             .bind(chunk)
             .bind(&chunk_keywords)
-            .execute(db)
+            .execute(&mut *tx)
             .await?;
         }
 
+        tx.commit().await?;
+
         tracing::info!(
             "GenerateEmbedding: item '{}' → {} chunks (keywords: {})",
             title,

crates/zclaw-saas/src/workers/update_last_used.rs

@@ -8,7 +8,8 @@ use super::Worker;
 
 #[derive(Debug, Serialize, Deserialize)]
 pub struct UpdateLastUsedArgs {
-    pub token_id: String,
+    /// token_hash 用于 WHERE 条件匹配
+    pub token_hash: String,
 }
 
 pub struct UpdateLastUsedWorker;
@@ -23,9 +24,9 @@ impl Worker for UpdateLastUsedWorker {
 
     async fn perform(&self, db: &PgPool, args: Self::Args) -> SaasResult<()> {
         let now = chrono::Utc::now().to_rfc3339();
-        sqlx::query("UPDATE api_tokens SET last_used_at = $1 WHERE id = $2")
+        sqlx::query("UPDATE api_tokens SET last_used_at = $1 WHERE token_hash = $2")
             .bind(&now)
-            .bind(&args.token_id)
+            .bind(&args.token_hash)
             .execute(db)
             .await?;
         Ok(())