实现 AI 图片理解能力测试

resources/mobile_daily_report.html

@@ -61,6 +61,13 @@
         font-size: 13px;
         line-height: 1.5;
       }
+      .record-note {
+        color: #485465;
+        font-weight: 650;
+      }
+      .overview {
+        margin-top: 2px;
+      }
       .avatar-grid {
         width: 58px;
         height: 58px;
@@ -422,7 +429,11 @@
         <div class="hero-top">
           <div>
             <h1>{{GROUP_NAME}}日报</h1>
-            <div class="sub">{{DATE_RANGE}}<br />{{RECORD_NOTE}}</div>
+            <div class="sub">
+              <div>{{DATE_RANGE}}</div>
+              <div class="record-note">{{RECORD_NOTE}}</div>
+              <div class="overview">{{OVERVIEW}}</div>
+            </div>
           </div>
           <div class="avatar-grid">{{HERO_AVATARS}}</div>
         </div>
@@ -479,7 +490,7 @@
       </section>
 
       <footer class="footer">
-        数据来源：微信群聊记录<br />
+        数据来源：WechatExplorer · 微信群聊记录<br />
         生成时间：{{GENERATED_AT}}<br />
         {{FOOTER_NOTE}}
       </footer>

src/main/group-report-service.ts

@@ -249,7 +249,8 @@ const renderReportHtml = async (request: GroupReportExportRequest): Promise<stri
     REPORT_TITLE: escapeHtml(`${metadata.groupName}日报`),
     GROUP_NAME: escapeHtml(metadata.groupName),
     DATE_RANGE: escapeHtml(metadata.dateRange),
-    RECORD_NOTE: escapeHtml(`${metadata.recordNote} ${report.overview}`.trim()),
+    RECORD_NOTE: escapeHtml(`基于 WechatExplorer 加载的 ${metadata.messageCount} 条记录`),
+    OVERVIEW: escapeHtml(report.overview),
     HERO_AVATARS: heroAvatars,
     MESSAGE_COUNT: String(metadata.messageCount),
     ACTIVE_USERS: String(metadata.activeUsers),

src/main/index.ts

@@ -27,7 +27,12 @@ import {
 } from './report-history-service'
 import type { GroupReportExportRequest } from '../shared/group-report'
 import type { SaveGeneratedReportRequest } from '../shared/report-history'
-import type { AIChatRequestOptions, AIProviderConfig, LegacyAIConfig } from '../shared/ai-provider'
+import type {
+  AIChatRequestOptions,
+  AIProviderConfig,
+  AIVisionTestRequest,
+  LegacyAIConfig
+} from '../shared/ai-provider'
 import { DatabaseKeyStore } from './database-key-store'
 import { ImageKeyConfigService } from './services/image-key-config-service'
 import { AIProviderService } from './services/ai-provider-service'
@@ -372,6 +377,9 @@ app.whenReady().then(async () => {
     aiProviderService.setDefault(providerId)
   )
   ipcMain.handle('ai:testProvider', (_, providerId: string) => aiProviderService.test(providerId))
+  ipcMain.handle('ai:testVision', (_, request: AIVisionTestRequest) =>
+    aiProviderService.testVision(request)
+  )
   ipcMain.handle('ai:migrateLegacy', (_, config: LegacyAIConfig) =>
     aiProviderService.migrateLegacy(config)
   )

src/main/services/ai-provider-service.ts

@@ -8,6 +8,8 @@ import type {
   AIProviderListResult,
   AIProviderSummary,
   AIRuntimeModelConfig,
+  AIVisionTestRequest,
+  AIVisionTestResult,
   LegacyAIConfig
 } from '../../shared/ai-provider'
 import { AIProviderKeyStore } from '../ai-provider-key-store'
@@ -18,7 +20,8 @@ interface AIProviderMetadataFile {
   providers: Array<Omit<AIProviderSummary, 'hasApiKey' | 'isDefault'>>
 }
 
-type AIMessage = { role: string; content: string }
+type AIMessagePart = { type: 'text'; text: string } | { type: 'image'; dataUrl: string }
+type AIMessage = { role: string; content: string | AIMessagePart[] }
 type AIRequestResult = {
   data: string
   usage?: { input?: number; output?: number; total?: number; estimated?: boolean }
@@ -151,7 +154,7 @@ export class AIProviderService {
   }
 
   async chat(
-    messages: AIMessage[],
+    messages: Array<{ role: string; content: string }>,
     options?: AIChatRequestOptions
   ): Promise<{
     success: boolean
@@ -166,6 +169,42 @@ export class AIProviderService {
     }
   }
 
+  async testVision(request: AIVisionTestRequest): Promise<AIVisionTestResult> {
+    const startedAt = Date.now()
+    const imageError = validateVisionImage(request.imageDataUrl)
+    if (imageError) return { success: false, code: 'INVALID_IMAGE', error: imageError }
+    if (!request.prompt.trim()) {
+      return { success: false, code: 'INVALID_IMAGE', error: '请填写图片识别提示词' }
+    }
+    try {
+      const resolved = this.resolveProvider(request)
+      const result = await requestProvider(resolved.provider, resolved.key, resolved.model, [
+        {
+          role: 'user',
+          content: [
+            { type: 'text', text: request.prompt.trim() },
+            { type: 'image', dataUrl: request.imageDataUrl }
+          ]
+        }
+      ])
+      if (!result.data.trim()) throw new Error('API 未返回识别内容')
+      this.markVisionCapability(resolved.provider.id, resolved.model)
+      const model = resolved.provider.models.find((item) => item.id === resolved.model)
+      return {
+        success: true,
+        providerName: resolved.provider.name,
+        modelId: resolved.model,
+        modelName: model?.name || resolved.model,
+        latencyMs: Date.now() - startedAt,
+        usage: result.usage,
+        answer: result.data
+      }
+    } catch (error) {
+      const failure = visionFailure(error)
+      return { success: false, ...failure, latencyMs: Date.now() - startedAt }
+    }
+  }
+
   private async request(
     messages: AIMessage[],
     options?: AIChatRequestOptions,
@@ -175,17 +214,25 @@ export class AIProviderService {
     usage?: { input?: number; output?: number; total?: number; estimated?: boolean }
   }> {
     if (options?.apiKey) return this.requestLegacy(messages, options)
+    const resolved = this.resolveProvider(options)
+    return requestProvider(resolved.provider, resolved.key, resolved.model, messages, testing)
+  }
+
+  private resolveProvider(options?: { providerId?: string; modelId?: string }): {
+    provider: AIProviderSummary
+    model: string
+    key: string
+  } {
     const list = this.list()
     const provider =
       list.providers.find((item) => item.id === options?.providerId) ||
       list.providers.find((item) => item.id === list.defaultProviderId)
     if (!provider) throw new Error('尚未配置 AI Provider')
     const model = options?.modelId || provider.defaultModel
+    if (!provider.models.some((item) => item.id === model)) throw new Error('当前模型不存在')
     const key = this.keyStore.get(provider.id).key || ''
     if (needsApiKey(provider) && !key) throw new Error('当前供应商尚未配置 API Key')
-    return provider.type === 'anthropic-messages'
-      ? requestAnthropic(provider, key, model, messages, testing)
-      : requestOpenAICompatible(provider, key, model, messages, testing)
+    return { provider, model, key }
   }
 
   private async requestLegacy(
@@ -215,6 +262,15 @@ export class AIProviderService {
     this.writeMetadata(data)
   }
 
+  private markVisionCapability(providerId: string, modelId: string): void {
+    const data = this.readMetadata()
+    const provider = data.providers.find((item) => item.id === providerId)
+    const model = provider?.models.find((item) => item.id === modelId)
+    if (!provider || !model || model.capabilities.vision) return
+    model.capabilities.vision = true
+    this.writeMetadata(data)
+  }
+
   private ensureEnvironmentMigration(): void {
     const data = this.readMetadata()
     if (data.providers.length) return
@@ -329,6 +385,51 @@ function buildHeaders(provider: AIProviderSummary, apiKey: string): Record<strin
   return headers
 }
 
+function requestProvider(
+  provider: AIProviderSummary,
+  apiKey: string,
+  model: string,
+  messages: AIMessage[],
+  testing = false
+): Promise<AIRequestResult> {
+  return provider.type === 'anthropic-messages'
+    ? requestAnthropic(provider, apiKey, model, messages, testing)
+    : requestOpenAICompatible(provider, apiKey, model, messages, testing)
+}
+
+function toOpenAIMessages(messages: AIMessage[]): Array<{ role: string; content: unknown }> {
+  return messages.map((message) => ({
+    role: message.role,
+    content:
+      typeof message.content === 'string'
+        ? message.content
+        : message.content.map((part) =>
+            part.type === 'text'
+              ? { type: 'text', text: part.text }
+              : { type: 'image_url', image_url: { url: part.dataUrl } }
+          )
+  }))
+}
+
+function toAnthropicMessages(messages: AIMessage[]): Array<{ role: string; content: unknown }> {
+  return messages
+    .filter((message) => message.role !== 'system')
+    .map((message) => ({
+      role: message.role,
+      content:
+        typeof message.content === 'string'
+          ? message.content
+          : message.content.map((part) => {
+              if (part.type === 'text') return { type: 'text', text: part.text }
+              const image = parseVisionImage(part.dataUrl)
+              return {
+                type: 'image',
+                source: { type: 'base64', media_type: image.mimeType, data: image.base64 }
+              }
+            })
+    }))
+}
+
 async function requestOpenAICompatible(
   provider: AIProviderSummary,
   apiKey: string,
@@ -346,7 +447,7 @@ async function requestOpenAICompatible(
       headers: buildHeaders(provider, apiKey),
       body: JSON.stringify({
         model,
-        messages,
+        messages: toOpenAIMessages(messages),
         temperature: provider.advanced.temperature,
         max_tokens: testing ? 8 : provider.advanced.maxTokens
       })
@@ -377,9 +478,16 @@ async function requestAnthropic(
 ): Promise<AIRequestResult> {
   const system = messages
     .filter((message) => message.role === 'system')
-    .map((message) => message.content)
+    .map((message) =>
+      typeof message.content === 'string'
+        ? message.content
+        : message.content
+            .filter((part) => part.type === 'text')
+            .map((part) => (part.type === 'text' ? part.text : ''))
+            .join('\n')
+    )
     .join('\n\n')
-  const anthropicMessages = messages.filter((message) => message.role !== 'system')
+  const anthropicMessages = toAnthropicMessages(messages)
   const headers = buildHeaders(provider, apiKey)
   if (!headers['anthropic-version']) headers['anthropic-version'] = '2023-06-01'
   const endpoint = provider.baseUrl.endsWith('/messages')
@@ -440,3 +548,35 @@ function safeAIError(error: unknown): string {
   const message = error instanceof Error ? error.message : String(error)
   return message.replace(/sk-[a-z0-9_-]+/gi, '***').slice(0, 300)
 }
+
+function parseVisionImage(dataUrl: string): { mimeType: string; base64: string; bytes: number } {
+  const match = /^data:(image\/(?:png|jpeg|webp));base64,([a-z0-9+/=]+)$/i.exec(dataUrl)
+  if (!match) throw new Error('图片格式不受支持，请选择 PNG、JPG、JPEG 或 WebP')
+  const bytes = Buffer.byteLength(match[2], 'base64')
+  return { mimeType: match[1].toLowerCase(), base64: match[2], bytes }
+}
+
+function validateVisionImage(dataUrl: string): string | undefined {
+  try {
+    const image = parseVisionImage(dataUrl)
+    if (!image.bytes) return '图片内容为空'
+    if (image.bytes > 10 * 1024 * 1024) return '图片不能超过 10 MB'
+    return undefined
+  } catch (error) {
+    return error instanceof Error ? error.message : '图片无法读取'
+  }
+}
+
+function visionFailure(error: unknown): {
+  code: 'VISION_UNSUPPORTED' | 'API_ERROR'
+  error: string
+} {
+  const message = safeAIError(error)
+  const unsupported =
+    /vision|multimodal|image[_ ]url|image input|image.*support|support.*image|图片.*不支持|不支持.*图片/i.test(
+      message
+    )
+  return unsupported
+    ? { code: 'VISION_UNSUPPORTED', error: '当前模型不支持图片理解' }
+    : { code: 'API_ERROR', error: message || 'API 返回错误' }
+}

src/preload/index.d.ts

@@ -26,6 +26,8 @@ import type {
   AIProviderConfig,
   AIProviderListResult,
   AIRuntimeModelConfig,
+  AIVisionTestRequest,
+  AIVisionTestResult,
   LegacyAIConfig
 } from '../shared/ai-provider'
 
@@ -108,6 +110,7 @@ declare global {
       deleteAIProvider: (providerId: string) => Promise<AIProviderListResult>
       setDefaultAIProvider: (providerId: string) => Promise<AIProviderListResult>
       testAIProvider: (providerId: string) => Promise<AIConnectionTestResult>
+      testAIVision: (request: AIVisionTestRequest) => Promise<AIVisionTestResult>
       migrateLegacyAIConfig: (config: LegacyAIConfig) => Promise<AIProviderListResult>
       copyImage: (base64String: string) => Promise<{ success: boolean; error?: string }>
       getVoiceData: (

src/preload/index.ts

@@ -2,7 +2,12 @@ import { contextBridge, ipcRenderer } from 'electron'
 import { electronAPI } from '@electron-toolkit/preload'
 import type { GroupReportExportRequest } from '../shared/group-report'
 import type { SaveGeneratedReportRequest } from '../shared/report-history'
-import type { AIChatRequestOptions, AIProviderConfig, LegacyAIConfig } from '../shared/ai-provider'
+import type {
+  AIChatRequestOptions,
+  AIProviderConfig,
+  AIVisionTestRequest,
+  LegacyAIConfig
+} from '../shared/ai-provider'
 
 // 渲染器的自定义 API
 const api = {
@@ -29,6 +34,7 @@ const api = {
   setDefaultAIProvider: (providerId: string) =>
     ipcRenderer.invoke('ai:setDefaultProvider', providerId),
   testAIProvider: (providerId: string) => ipcRenderer.invoke('ai:testProvider', providerId),
+  testAIVision: (request: AIVisionTestRequest) => ipcRenderer.invoke('ai:testVision', request),
   migrateLegacyAIConfig: (config: LegacyAIConfig) => ipcRenderer.invoke('ai:migrateLegacy', config),
   copyImage: (base64String) => ipcRenderer.invoke('copy-image', base64String),
   getVoiceData: (sessionId: string, localId: number, createTime: number, svrId?: string | number) =>

src/renderer/src/assets/main.css

@@ -3429,6 +3429,199 @@ body {
   }
 }
 
+.ai-vision-test {
+  display: grid;
+  gap: 16px;
+}
+
+.ai-vision-test > header,
+.ai-vision-test > footer,
+.ai-vision-model {
+  display: flex;
+  align-items: center;
+  justify-content: space-between;
+  gap: 16px;
+}
+
+.ai-vision-test h2,
+.ai-vision-test h3,
+.ai-vision-test p {
+  margin: 0;
+}
+
+.ai-vision-test h2 {
+  color: #202724;
+  font-size: 15px;
+}
+
+.ai-vision-test header h3 {
+  margin-top: 12px;
+  color: #35403b;
+  font-size: 13px;
+}
+
+.ai-vision-test header p,
+.ai-vision-model,
+.ai-vision-upload small {
+  margin-top: 4px;
+  color: #66706b;
+  font-size: 12px;
+}
+
+.ai-vision-capability {
+  border-radius: 999px;
+  padding: 5px 9px;
+  background: #f1f3f2;
+  color: #66706b;
+  white-space: nowrap;
+  font-size: 12px;
+}
+
+.ai-vision-capability.supported {
+  background: #eaf5f1;
+  color: #2e8b68;
+}
+
+.ai-vision-model {
+  justify-content: flex-start;
+  margin: 0;
+}
+
+.ai-vision-upload {
+  display: flex;
+  min-height: 112px;
+  align-items: center;
+  justify-content: center;
+  gap: 14px;
+  border: 1px dashed #bfc9c4;
+  border-radius: 10px;
+  padding: 14px;
+  background: #f7f9f8;
+  color: #35403b;
+  text-align: center;
+  cursor: pointer;
+}
+
+.ai-vision-upload:hover {
+  border-color: #247a63;
+  background: #f2f8f5;
+}
+
+.ai-vision-upload input {
+  display: none;
+}
+
+.ai-vision-upload.has-image {
+  justify-content: flex-start;
+  text-align: left;
+}
+
+.ai-vision-upload img {
+  width: 112px;
+  height: 82px;
+  flex: 0 0 auto;
+  border-radius: 8px;
+  object-fit: cover;
+}
+
+.ai-vision-upload strong,
+.ai-vision-upload small {
+  display: block;
+}
+
+.ai-vision-prompt {
+  display: grid;
+  gap: 7px;
+  color: #46514c;
+  font-size: 12px;
+  font-weight: 600;
+}
+
+.ai-vision-prompt textarea {
+  width: 100%;
+  border: 1px solid #d5ddda;
+  border-radius: 8px;
+  outline: none;
+  padding: 10px;
+  resize: vertical;
+  background: #f4f7f5;
+  color: #202724;
+  font: 13px/1.6 var(--wxex-font);
+}
+
+.ai-vision-prompt textarea:focus {
+  border-color: #247a63;
+  box-shadow: 0 0 0 2px rgba(36, 122, 99, 0.08);
+}
+
+.ai-vision-privacy {
+  color: #66706b;
+  font-size: 12px;
+  line-height: 1.6;
+}
+
+.ai-vision-error {
+  border-left: 3px solid #c85a5a;
+  padding: 9px 11px;
+  background: #fff2f2;
+  color: #a84444;
+  font-size: 12px;
+}
+
+.ai-vision-result {
+  display: grid;
+  gap: 12px;
+  border: 1px solid #dce7e2;
+  border-radius: 9px;
+  padding: 14px;
+  background: #f7faf9;
+}
+
+.ai-vision-result dl {
+  display: grid;
+  grid-template-columns: repeat(3, minmax(0, 1fr));
+  gap: 12px;
+  margin: 0;
+}
+
+.ai-vision-result dt {
+  color: #929a96;
+  font-size: 11px;
+}
+
+.ai-vision-result dd {
+  margin: 4px 0 0;
+  color: #35403b;
+  font-size: 12px;
+  font-weight: 600;
+}
+
+.ai-vision-result > p {
+  white-space: pre-wrap;
+  color: #3d4742;
+  font-size: 13px;
+  line-height: 1.7;
+}
+
+.ai-vision-test > footer {
+  justify-content: flex-end;
+}
+
+.ai-vision-test > footer button {
+  min-height: 34px;
+  border: 1px solid #d5ddda;
+  border-radius: 7px;
+  padding: 0 13px;
+  background: #fff;
+  color: #46514c;
+  cursor: pointer;
+}
+
+.ai-vision-test > footer button:disabled {
+  cursor: not-allowed;
+  opacity: 0.5;
+}
+
 .report-density-options button:disabled:hover {
   border-color: var(--wxex-border);
   background: var(--wxex-bg-elevated);

src/renderer/src/features/settings/ai-model/AIImageUnderstandingTest.tsx

@@ -0,0 +1,121 @@
+import type { AIProviderSummary, AIRuntimeModelConfig } from '../../../../../shared/ai-provider'
+import type { AIVisionTestState } from './types'
+
+export function AIImageUnderstandingTest({
+  runtime,
+  provider,
+  state,
+  onSelectImage,
+  onPromptChange,
+  onTest,
+  onClear
+}: {
+  runtime: AIRuntimeModelConfig | null
+  provider?: AIProviderSummary
+  state: AIVisionTestState
+  onSelectImage: (file: File) => void
+  onPromptChange: (prompt: string) => void
+  onTest: () => void
+  onClear: () => void
+}): React.ReactElement {
+  const model = provider?.models.find((item) => item.id === runtime?.model)
+  const testing = state.status === 'testing'
+  const result = state.result
+  return (
+    <section className="settings-card ai-vision-test">
+      <header>
+        <div>
+          <h2>AI 能力测试</h2>
+          <h3>图片理解测试</h3>
+          <p>上传图片验证当前模型是否支持视觉理解。</p>
+        </div>
+        <span className={`ai-vision-capability ${model?.capabilities.vision ? 'supported' : ''}`}>
+          图片理解 {model?.capabilities.vision ? '✓' : '待验证'}
+        </span>
+      </header>
+
+      <div className="ai-vision-model">
+        <span>当前供应商：{runtime?.providerName || '尚未配置'}</span>
+        <span>当前模型：{runtime?.modelName || '尚未选择'}</span>
+      </div>
+
+      <label className={`ai-vision-upload ${state.image ? 'has-image' : ''}`}>
+        <input
+          type="file"
+          accept=".png,.jpg,.jpeg,.webp,image/png,image/jpeg,image/webp"
+          onChange={(event) => {
+            const file = event.currentTarget.files?.[0]
+            if (file) onSelectImage(file)
+            event.currentTarget.value = ''
+          }}
+        />
+        {state.image ? (
+          <>
+            <img src={state.image.dataUrl} alt="图片理解测试预览" />
+            <div>
+              <strong>{state.image.fileName}</strong>
+              <small>{formatFileSize(state.image.size)} · 仅保存在内存中</small>
+            </div>
+          </>
+        ) : (
+          <div>
+            <strong>{state.status === 'reading' ? '正在读取图片…' : '选择本地图片'}</strong>
+            <small>支持 PNG、JPG、JPEG、WebP，最大 10 MB</small>
+          </div>
+        )}
+      </label>
+
+      <label className="ai-vision-prompt">
+        识别提示词
+        <textarea
+          value={state.prompt}
+          onChange={(event) => onPromptChange(event.target.value)}
+          rows={3}
+        />
+      </label>
+
+      <p className="ai-vision-privacy">
+        图片只会发送给你配置的 AI 服务，不会上传到 WechatExplorer 的其他服务器，也不会写入本地缓存。
+      </p>
+
+      {state.error ? <p className="ai-vision-error">{state.error}</p> : null}
+      {result?.success ? (
+        <div className="ai-vision-result">
+          <h3>识别结果</h3>
+          <dl>
+            <div>
+              <dt>模型</dt>
+              <dd>{result.modelName || result.modelId}</dd>
+            </div>
+            <div>
+              <dt>耗时</dt>
+              <dd>{result.latencyMs ?? 0} ms</dd>
+            </div>
+            <div>
+              <dt>Token</dt>
+              <dd>{result.usage?.total ?? 'API 未返回'}</dd>
+            </div>
+          </dl>
+          <p>{result.answer}</p>
+        </div>
+      ) : null}
+
+      <footer>
+        {state.image ? <button onClick={onClear}>移除图片</button> : null}
+        <button
+          className="database-key-primary"
+          disabled={!runtime?.configured || !state.image || testing || !state.prompt.trim()}
+          onClick={onTest}
+        >
+          {testing ? '识别中…' : '开始识别'}
+        </button>
+      </footer>
+    </section>
+  )
+}
+
+function formatFileSize(bytes: number): string {
+  return bytes < 1024 * 1024
+    ? `${Math.max(1, Math.round(bytes / 1024))} KB`
+    : `${(bytes / 1024 / 1024).toFixed(1)} MB`
+}

src/renderer/src/features/settings/ai-model/aiModelSettingsReducer.ts

@@ -1,4 +1,5 @@
 import type { AIModelSettingsAction, AIModelSettingsState } from './types'
+import { DEFAULT_VISION_PROMPT } from './types'
 
 export const initialAIModelSettingsState: AIModelSettingsState = {
   loading: true,
@@ -6,7 +7,8 @@ export const initialAIModelSettingsState: AIModelSettingsState = {
   providers: [],
   runtime: null,
   editor: null,
-  presetId: 'deepseek'
+  presetId: 'deepseek',
+  visionTest: { status: 'idle', prompt: DEFAULT_VISION_PROMPT }
 }
 
 export function aiModelSettingsReducer(
@@ -42,6 +44,49 @@ export function aiModelSettingsReducer(
       return { ...state, saving: true, error: undefined }
     case 'TEST_START':
       return { ...state, testingId: action.providerId, error: undefined }
+    case 'VISION_READING':
+      return {
+        ...state,
+        visionTest: { ...state.visionTest, status: 'reading', result: undefined, error: undefined }
+      }
+    case 'VISION_READY':
+      return {
+        ...state,
+        visionTest: {
+          ...state.visionTest,
+          status: 'ready',
+          image: action.image,
+          result: undefined,
+          error: undefined
+        }
+      }
+    case 'VISION_PROMPT':
+      return { ...state, visionTest: { ...state.visionTest, prompt: action.prompt } }
+    case 'VISION_TEST_START':
+      return {
+        ...state,
+        visionTest: { ...state.visionTest, status: 'testing', result: undefined, error: undefined }
+      }
+    case 'VISION_RESULT':
+      return {
+        ...state,
+        visionTest: {
+          ...state.visionTest,
+          status: action.result.success ? 'success' : 'error',
+          result: action.result,
+          error: action.result.error
+        }
+      }
+    case 'VISION_ERROR':
+      return {
+        ...state,
+        visionTest: { ...state.visionTest, status: 'error', result: undefined, error: action.error }
+      }
+    case 'VISION_CLEAR':
+      return {
+        ...state,
+        visionTest: { status: 'idle', prompt: state.visionTest.prompt }
+      }
     default:
       return state
   }

src/renderer/src/features/settings/ai-model/types.ts

@@ -1,9 +1,26 @@
 import type {
   AIProviderConfig,
   AIProviderSummary,
-  AIRuntimeModelConfig
+  AIRuntimeModelConfig,
+  AIVisionTestResult
 } from '../../../../../shared/ai-provider'
 
+export const DEFAULT_VISION_PROMPT =
+  '请描述这张图片中的主要内容，包括物体、场景、文字信息以及你能观察到的细节。'
+
+export interface AIVisionTestState {
+  status: 'idle' | 'reading' | 'ready' | 'testing' | 'success' | 'error'
+  prompt: string
+  image?: {
+    dataUrl: string
+    fileName: string
+    mimeType: string
+    size: number
+  }
+  result?: AIVisionTestResult
+  error?: string
+}
+
 export interface AIModelSettingsState {
   loading: boolean
   saving: boolean
@@ -13,6 +30,7 @@ export interface AIModelSettingsState {
   originalProviderId?: string
   presetId: string
   testingId?: string
+  visionTest: AIVisionTestState
   error?: string
 }
 
@@ -29,6 +47,16 @@ export type AIModelSettingsAction =
   | { type: 'EDIT'; editor: AIProviderConfig }
   | { type: 'SAVE_START' }
   | { type: 'TEST_START'; providerId: string }
+  | { type: 'VISION_READING' }
+  | {
+      type: 'VISION_READY'
+      image: NonNullable<AIVisionTestState['image']>
+    }
+  | { type: 'VISION_PROMPT'; prompt: string }
+  | { type: 'VISION_TEST_START' }
+  | { type: 'VISION_RESULT'; result: AIVisionTestResult }
+  | { type: 'VISION_ERROR'; error: string }
+  | { type: 'VISION_CLEAR' }
 
 export interface AIModelSettingsController {
   state: AIModelSettingsState
@@ -41,4 +69,8 @@ export interface AIModelSettingsController {
   remove: (providerId: string) => Promise<void>
   setDefault: (providerId: string) => Promise<void>
   test: (providerId: string) => Promise<void>
+  selectVisionImage: (file: File) => Promise<void>
+  setVisionPrompt: (prompt: string) => void
+  runVisionTest: () => Promise<void>
+  clearVisionImage: () => void
 }

src/renderer/src/features/settings/ai-model/useAIModelSettingsController.ts

@@ -108,6 +108,71 @@ export function useAIModelSettingsController({
     [onNotice, refresh]
   )
 
+  const selectVisionImage = useCallback(async (file: File): Promise<void> => {
+    const extension = file.name.split('.').pop()?.toLowerCase()
+    const inferredType =
+      extension === 'png'
+        ? 'image/png'
+        : extension === 'webp'
+          ? 'image/webp'
+          : extension === 'jpg' || extension === 'jpeg'
+            ? 'image/jpeg'
+            : ''
+    const mimeType = file.type === 'image/jpg' ? 'image/jpeg' : file.type || inferredType
+    const supportedTypes = new Set(['image/png', 'image/jpeg', 'image/webp'])
+    if (!supportedTypes.has(mimeType)) {
+      return dispatch({ type: 'VISION_ERROR', error: '请选择 PNG、JPG、JPEG 或 WebP 图片' })
+    }
+    if (!file.size || file.size > 10 * 1024 * 1024) {
+      return dispatch({ type: 'VISION_ERROR', error: '图片大小必须在 10 MB 以内' })
+    }
+    dispatch({ type: 'VISION_READING' })
+    try {
+      const rawDataUrl = await readFileAsDataUrl(file)
+      const dataUrl = rawDataUrl.replace(/^data:[^;]*;/, `data:${mimeType};`)
+      dispatch({
+        type: 'VISION_READY',
+        image: { dataUrl, fileName: file.name, mimeType, size: file.size }
+      })
+    } catch {
+      dispatch({ type: 'VISION_ERROR', error: '图片无法读取，请重新选择' })
+    }
+  }, [])
+
+  const setVisionPrompt = useCallback(
+    (prompt: string) => dispatch({ type: 'VISION_PROMPT', prompt }),
+    []
+  )
+
+  const runVisionTest = useCallback(async (): Promise<void> => {
+    const { runtime, visionTest } = state
+    if (!runtime?.configured || !runtime.providerId || !runtime.model) {
+      return dispatch({ type: 'VISION_ERROR', error: '请先配置可用的默认 AI 模型' })
+    }
+    if (!visionTest.image) return dispatch({ type: 'VISION_ERROR', error: '请先选择测试图片' })
+    if (!visionTest.prompt.trim()) {
+      return dispatch({ type: 'VISION_ERROR', error: '请填写图片识别提示词' })
+    }
+    dispatch({ type: 'VISION_TEST_START' })
+    try {
+      const result = await window.api.testAIVision({
+        providerId: runtime.providerId,
+        modelId: runtime.model,
+        prompt: visionTest.prompt,
+        imageDataUrl: visionTest.image.dataUrl
+      })
+      dispatch({ type: 'VISION_RESULT', result })
+      if (result.success) {
+        await refresh()
+        onNotice('图片理解测试成功，已更新模型能力')
+      }
+    } catch {
+      dispatch({ type: 'VISION_ERROR', error: '图片理解测试调用失败，请稍后重试' })
+    }
+  }, [onNotice, refresh, state])
+
+  const clearVisionImage = useCallback(() => dispatch({ type: 'VISION_CLEAR' }), [])
+
   return {
     state,
     openNew,
@@ -118,6 +183,23 @@ export function useAIModelSettingsController({
     save,
     remove,
     setDefault,
-    test
+    test,
+    selectVisionImage,
+    setVisionPrompt,
+    runVisionTest,
+    clearVisionImage
   }
 }
+
+function readFileAsDataUrl(file: File): Promise<string> {
+  return new Promise((resolve, reject) => {
+    const reader = new FileReader()
+    reader.addEventListener('load', () =>
+      typeof reader.result === 'string'
+        ? resolve(reader.result)
+        : reject(new Error('invalid image'))
+    )
+    reader.addEventListener('error', () => reject(reader.error || new Error('read failed')))
+    reader.readAsDataURL(file)
+  })
+}

src/renderer/src/features/settings/pages/AIModelPage.tsx

@@ -1,6 +1,7 @@
 import type { AIRuntimeModelConfig } from '../../../../../shared/ai-provider'
 import { AIProviderCard } from '../ai-model/AIProviderCard'
 import { AIProviderEditor } from '../ai-model/AIProviderEditor'
+import { AIImageUnderstandingTest } from '../ai-model/AIImageUnderstandingTest'
 import { useAIModelSettingsController } from '../ai-model/useAIModelSettingsController'
 
 export function AIModelPage({
@@ -12,6 +13,9 @@ export function AIModelPage({
 }): React.ReactElement {
   const controller = useAIModelSettingsController({ onRuntimeChange, onNotice })
   const runtime = controller.state.runtime
+  const defaultProvider = controller.state.providers.find(
+    (provider) => provider.id === runtime?.providerId
+  )
   return (
     <div className="settings-page ai-model-page">
       <header className="settings-page-header">
@@ -35,6 +39,15 @@ export function AIModelPage({
               {runtime?.configured ? '可用' : '未配置'}
             </span>
           </section>
+          <AIImageUnderstandingTest
+            runtime={runtime}
+            provider={defaultProvider}
+            state={controller.state.visionTest}
+            onSelectImage={(file) => void controller.selectVisionImage(file)}
+            onPromptChange={controller.setVisionPrompt}
+            onTest={() => void controller.runVisionTest()}
+            onClear={controller.clearVisionImage}
+          />
           {controller.state.error ? (
             <p className="ai-model-page-error">{controller.state.error}</p>
           ) : null}

src/renderer/src/utils/group-report.ts

@@ -282,7 +282,7 @@ export const buildGroupReportInput = (
     activeUsers: speakerCounts.size,
     timeSpan,
     generatedAt: new Date().toLocaleString('zh-CN', { hour12: false }),
-    recordNote: `基于当前已加载的 ${rows.length} 条记录`,
+    recordNote: `基于 WechatExplorer 加载的 ${rows.length} 条记录`,
     footerNote: '基于已读取聊天记录生成；图片、表情等未解析内容仅按类型统计。',
     heroParticipants: topSpeakers.slice(0, 4).map((speaker) => speaker.name),
     avatars

src/shared/ai-provider.ts

@@ -88,3 +88,27 @@ export interface AIConnectionTestResult {
   error?: string
   latencyMs?: number
 }
+
+export interface AIVisionTestRequest {
+  providerId?: string
+  modelId?: string
+  prompt: string
+  imageDataUrl: string
+}
+
+export interface AIVisionTestResult {
+  success: boolean
+  providerName?: string
+  modelId?: string
+  modelName?: string
+  latencyMs?: number
+  usage?: {
+    input?: number
+    output?: number
+    total?: number
+    estimated?: boolean
+  }
+  answer?: string
+  code?: 'INVALID_IMAGE' | 'VISION_UNSUPPORTED' | 'API_ERROR'
+  error?: string
+}