fix(ai): drop queued stream text after stop

Stop generation now cancels the text batcher instead of flushing it, and queued deltas are ignored once the assistant message is no longer streaming.

src/stores/aiChat.ts

@@ -44,7 +44,7 @@ import {
   toSerializableContentBlocks,
 } from './aiChatContentBlocks'
 import { createToolLifecycleTracker, type ToolStatus } from './aiToolLifecycle'
-import { createAIStreamTextBatcher } from './aiChatStreamBatcher'
+import { applyQueuedStreamTextDeltas, createAIStreamTextBatcher } from './aiChatStreamBatcher'
 
 export type { ToolStatus } from './aiToolLifecycle'
 
@@ -666,6 +666,7 @@ export const useAIChatStore = defineStore('aiChatRuntime', () => {
     completeTurn: (hadToolCalls: boolean) => void
     settleProcessDuration: () => void
     flushPendingText: () => void
+    discardPendingText: () => void
   }
 
   function createStreamBlockHelpers(targetBuffer: AIChatBuffer, getAiMessageIndex: () => number): StreamBlockHelpers {
@@ -705,35 +706,16 @@ export const useAIChatStore = defineStore('aiChatRuntime', () => {
     const streamTextBatcher = createAIStreamTextBatcher((deltas) => {
       const idx = getAiMessageIndex()
       const message = targetBuffer.messages[idx]
-      const blocks = message.contentBlocks || []
-      let content = message.content
-      let changed = false
-
-      for (const delta of deltas) {
-        const lastBlock = blocks[blocks.length - 1]
-        if (delta.type === 'content') {
-          if (delta.content.trim().length === 0 && (!lastBlock || lastBlock.type !== 'text')) continue
-          if (lastBlock?.type === 'text') lastBlock.text += delta.content
-          else blocks.push({ type: 'text', text: delta.content })
-          content += delta.content
-          changed = true
-          continue
-        }
-
-        const thinkTag = delta.thinkTag || 'think'
-        if (lastBlock?.type === 'think' && lastBlock.tag === thinkTag) {
-          lastBlock.text += delta.content
-          changed = true
-        } else if (delta.content.trim().length > 0) {
-          blocks.push({ type: 'think', tag: thinkTag, text: delta.content })
-          changed = true
-        }
-      }
-
-      if (changed) applyAIMessageUpdates({ contentBlocks: [...blocks], content })
+      const applied = applyQueuedStreamTextDeltas(message, deltas)
+      if (!applied) return
+      applyAIMessageUpdates({
+        content: applied.content,
+        contentBlocks: applied.contentBlocks as ContentBlock[],
+      })
     })
 
     const flushPendingText = () => streamTextBatcher.flush()
+    const discardPendingText = () => streamTextBatcher.cancel()
 
     const updateAIMessage = (updates: Partial<ChatMessage>) => {
       flushPendingText()
@@ -939,6 +921,7 @@ export const useAIChatStore = defineStore('aiChatRuntime', () => {
       completeTurn,
       settleProcessDuration,
       flushPendingText,
+      discardPendingText,
     }
   }
 
@@ -1067,6 +1050,7 @@ export const useAIChatStore = defineStore('aiChatRuntime', () => {
         completeTurn,
         settleProcessDuration,
         flushPendingText,
+        discardPendingText,
       } = createStreamBlockHelpers(targetBuffer, () => aiMessageIndex)
 
       const currentAssistantId = targetBuffer.assistantId ?? getDefaultGeneralAssistantId(state.locale)
@@ -1314,11 +1298,12 @@ export const useAIChatStore = defineStore('aiChatRuntime', () => {
       setActiveTaskMeta(chatKey, content, agentReqId, resolvedAIChatId)
 
       const result = await agentPromise
-      flushPendingText()
       if (state.isAborted) {
+        discardPendingText()
         clearActiveTask(chatKey, agentReqId)
         return { success: false, reason: 'aborted' }
       }
+      flushPendingText()
 
       if (thisRequestId !== state.currentRequestId) {
         clearActiveTask(chatKey, agentReqId)
@@ -1694,6 +1679,7 @@ export const useAIChatStore = defineStore('aiChatRuntime', () => {
       completeTurn,
       settleProcessDuration,
       flushPendingText,
+      discardPendingText,
     } = createStreamBlockHelpers(targetBuffer, () => aiMessageIndex)
 
     try {
@@ -1884,12 +1870,13 @@ export const useAIChatStore = defineStore('aiChatRuntime', () => {
       setActiveTaskMeta(chatKey, content, agentReqId, state.currentAIChatId)
 
       const result = await agentPromise
-      flushPendingText()
       if (state.isAborted) {
+        discardPendingText()
         restoreOriginal()
         clearActiveTask(chatKey, agentReqId)
         return { success: false, reason: 'aborted' }
       }
+      flushPendingText()
       if (thisRequestId !== state.currentRequestId) {
         restoreOriginal()
         clearActiveTask(chatKey, agentReqId)

src/stores/aiChatStreamBatcher.test.ts

@@ -1,6 +1,6 @@
 import assert from 'node:assert/strict'
 import test from 'node:test'
-import { createAIStreamTextBatcher, type AIStreamTextDelta } from './aiChatStreamBatcher'
+import { applyQueuedStreamTextDeltas, createAIStreamTextBatcher, type AIStreamTextDelta } from './aiChatStreamBatcher'
 
 function createManualScheduler() {
   let nextHandle = 1
@@ -69,6 +69,36 @@ test('cancels the pending frame without applying stale deltas', () => {
   batcher.push({ type: 'content', content: 'stale' })
   batcher.cancel()
   frame.runFrame()
+  batcher.flush()
 
   assert.deepEqual(applied, [])
 })
+
+test('does not append queued text after the assistant message has stopped', () => {
+  const message = {
+    isStreaming: false,
+    content: 'hello\n\n_（已停止生成）_',
+    contentBlocks: [{ type: 'text', text: 'hello' }],
+  }
+
+  const applied = applyQueuedStreamTextDeltas(message, [{ type: 'content', content: ' world' }])
+
+  assert.equal(applied, null)
+  assert.equal(message.content, 'hello\n\n_（已停止生成）_')
+  assert.deepEqual(message.contentBlocks, [{ type: 'text', text: 'hello' }])
+})
+
+test('appends queued text while the assistant message is still streaming', () => {
+  const message = {
+    isStreaming: true,
+    content: 'hel',
+    contentBlocks: [{ type: 'text', text: 'hel' }],
+  }
+
+  const applied = applyQueuedStreamTextDeltas(message, [{ type: 'content', content: 'lo' }])
+
+  assert.deepEqual(applied, {
+    content: 'hello',
+    contentBlocks: [{ type: 'text', text: 'hello' }],
+  })
+})

src/stores/aiChatStreamBatcher.ts

@@ -2,6 +2,47 @@ export type AIStreamTextDelta =
   | { type: 'content'; content: string }
   | { type: 'think'; content: string; thinkTag?: string }
 
+export interface StreamTextTarget {
+  isStreaming?: boolean
+  content: string
+  contentBlocks?: Array<{ type: string; text?: string; tag?: string }>
+}
+
+export function applyQueuedStreamTextDeltas(
+  message: StreamTextTarget,
+  deltas: AIStreamTextDelta[]
+): { content: string; contentBlocks: NonNullable<StreamTextTarget['contentBlocks']> } | null {
+  if (!message.isStreaming) return null
+
+  const blocks = message.contentBlocks ? [...message.contentBlocks] : []
+  let content = message.content
+  let changed = false
+
+  for (const delta of deltas) {
+    const lastBlock = blocks[blocks.length - 1]
+    if (delta.type === 'content') {
+      if (delta.content.trim().length === 0 && lastBlock?.type !== 'text') continue
+      if (lastBlock?.type === 'text') lastBlock.text = (lastBlock.text || '') + delta.content
+      else blocks.push({ type: 'text', text: delta.content })
+      content += delta.content
+      changed = true
+      continue
+    }
+
+    const thinkTag = delta.thinkTag || 'think'
+    if (lastBlock?.type === 'think' && lastBlock.tag === thinkTag) {
+      lastBlock.text = (lastBlock.text || '') + delta.content
+      changed = true
+    } else if (delta.content.trim().length > 0) {
+      blocks.push({ type: 'think', tag: thinkTag, text: delta.content })
+      changed = true
+    }
+  }
+
+  if (!changed) return { content: message.content, contentBlocks: message.contentBlocks || [] }
+  return { content, contentBlocks: blocks }
+}
+
 interface FrameScheduler {
   request: (callback: () => void) => number
   cancel: (handle: number) => void