Merge PR #672: 支持账号批量配置与可用模型管理

frontend/src/features/providers/composables/__tests__/useUpstreamModelsCache.spec.ts

@@ -0,0 +1,71 @@
+import { beforeEach, describe, expect, it, vi } from 'vitest'
+
+const adminApiMocks = vi.hoisted(() => ({
+  queryProviderModels: vi.fn(),
+  queryProviderModelsForKeys: vi.fn(),
+}))
+
+vi.mock('@/api/admin', () => ({ adminApi: adminApiMocks }))
+
+import { useUpstreamModelsCache } from '../useUpstreamModelsCache'
+
+function response(modelId: string) {
+  return {
+    success: true,
+    data: { models: [{ id: modelId }] },
+    provider: { id: 'provider-1', name: 'Provider', display_name: 'Provider' },
+  }
+}
+
+function deferred<T>() {
+  let resolve!: (value: T) => void
+  const promise = new Promise<T>((done) => { resolve = done })
+  return { promise, resolve }
+}
+
+describe('useUpstreamModelsCache', () => {
+  beforeEach(() => {
+    adminApiMocks.queryProviderModels.mockReset()
+    adminApiMocks.queryProviderModelsForKeys.mockReset()
+  })
+
+  it('deduplicates equivalent multi-key model requests', async () => {
+    const request = deferred<ReturnType<typeof response>>()
+    adminApiMocks.queryProviderModelsForKeys.mockReturnValue(request.promise)
+    const { fetchModelsForKeys } = useUpstreamModelsCache()
+
+    const first = fetchModelsForKeys('provider-1', ['key-b', 'key-a', 'key-a'])
+    const second = fetchModelsForKeys('provider-1', ['key-a', 'key-b'])
+    expect(adminApiMocks.queryProviderModelsForKeys).toHaveBeenCalledTimes(1)
+    expect(adminApiMocks.queryProviderModelsForKeys).toHaveBeenCalledWith(
+      'provider-1',
+      ['key-a', 'key-b'],
+      false,
+    )
+
+    request.resolve(response('gpt-5.6-sol'))
+    await expect(first).resolves.toMatchObject({ models: [{ id: 'gpt-5.6-sol' }] })
+    await expect(second).resolves.toMatchObject({ models: [{ id: 'gpt-5.6-sol' }] })
+  })
+
+  it('keeps the loading state owned by the latest forced request', async () => {
+    const firstRequest = deferred<ReturnType<typeof response>>()
+    const forcedRequest = deferred<ReturnType<typeof response>>()
+    adminApiMocks.queryProviderModels
+      .mockReturnValueOnce(firstRequest.promise)
+      .mockReturnValueOnce(forcedRequest.promise)
+    const { fetchModels, isLoading } = useUpstreamModelsCache()
+
+    const first = fetchModels('provider-1', 'key-a')
+    const forced = fetchModels('provider-1', 'key-a', true)
+    expect(isLoading('provider-1', 'key-a')).toBe(true)
+
+    firstRequest.resolve(response('gpt-5.6-sol'))
+    await first
+    expect(isLoading('provider-1', 'key-a')).toBe(true)
+
+    forcedRequest.resolve(response('gpt-5.6-luna'))
+    await forced
+    expect(isLoading('provider-1', 'key-a')).toBe(false)
+  })
+})

frontend/src/features/providers/composables/useUpstreamModelsCache.ts

@@ -5,7 +5,7 @@
  */
 import { ref } from 'vue'
 import { isAxiosError } from 'axios'
-import { adminApi } from '@/api/admin'
+import { adminApi, type ProviderModelsQueryResponse } from '@/api/admin'
 import { parseUpstreamModelError } from '@/utils/errorParser'
 import type { UpstreamModel } from '@/api/endpoints/types'
 
@@ -15,6 +15,8 @@ type FetchResult = { models: UpstreamModel[]; error?: string; warning?: string;
 
 // 进行中的请求（用于去重并发请求）
 const pendingRequests = new Map<string, Promise<FetchResult>>()
+const activeRequestIds = new Map<string, number>()
+let nextRequestId = 0
 
 // 请求状态
 const loadingMap = ref<Map<string, boolean>>(new Map())
@@ -26,6 +28,57 @@ function getRequestKey(providerId: string, apiKeyId?: string): string {
   return apiKeyId ? `${providerId}:${apiKeyId}` : providerId
 }
 
+function getBatchRequestKey(providerId: string, apiKeyIds: string[]): string {
+  return `${providerId}:batch:${JSON.stringify([...new Set(apiKeyIds)].sort())}`
+}
+
+function providerModelsFetchResult(response: ProviderModelsQueryResponse): FetchResult {
+  if (response.success && response.data?.models) {
+    const partialWarning = response.data.warning ?? response.data.error
+    return {
+      models: response.data.models,
+      warning: partialWarning ? parseUpstreamModelError(partialWarning) : undefined,
+      fromCache: response.data.from_cache,
+    }
+  }
+  const rawError = response.data?.error || response.data?.warning || '获取上游模型失败'
+  return { models: [], error: parseUpstreamModelError(rawError) }
+}
+
+function fetchProviderModels(
+  requestKey: string,
+  forceRefresh: boolean,
+  request: () => Promise<ProviderModelsQueryResponse>,
+): Promise<FetchResult> {
+  if (!forceRefresh && pendingRequests.has(requestKey)) {
+    // eslint-disable-next-line @typescript-eslint/no-non-null-assertion
+    return pendingRequests.get(requestKey)!
+  }
+
+  const requestId = ++nextRequestId
+  activeRequestIds.set(requestKey, requestId)
+  loadingMap.value.set(requestKey, true)
+  const requestPromise = (async (): Promise<FetchResult> => {
+    try {
+      return providerModelsFetchResult(await request())
+    } catch (err: unknown) {
+      const rawError = isAxiosError(err)
+        ? (err.response?.data?.detail ?? err.message)
+        : (err instanceof Error ? err.message : String(err))
+      return { models: [], error: parseUpstreamModelError(rawError || '获取上游模型失败') }
+    } finally {
+      if (activeRequestIds.get(requestKey) === requestId) {
+        loadingMap.value.set(requestKey, false)
+        pendingRequests.delete(requestKey)
+        activeRequestIds.delete(requestKey)
+      }
+    }
+  })()
+
+  pendingRequests.set(requestKey, requestPromise)
+  return requestPromise
+}
+
 export function useUpstreamModelsCache() {
   /**
    * 获取上游模型列表
@@ -40,41 +93,32 @@ export function useUpstreamModelsCache() {
     forceRefresh = false
   ): Promise<FetchResult> {
     const requestKey = getRequestKey(providerId, apiKeyId)
+    return fetchProviderModels(
+      requestKey,
+      forceRefresh,
+      () => adminApi.queryProviderModels(providerId, apiKeyId, forceRefresh),
+    )
+  }
 
-    // 强制刷新时不复用进行中的请求
-    if (!forceRefresh && pendingRequests.has(requestKey)) {
-      // eslint-disable-next-line @typescript-eslint/no-non-null-assertion
-      return pendingRequests.get(requestKey)!
+  async function fetchModelsForKeys(
+    providerId: string,
+    apiKeyIds: string[],
+    forceRefresh = false
+  ): Promise<FetchResult> {
+    const normalizedKeyIds = [...new Set(apiKeyIds.map(id => id.trim()).filter(Boolean))].sort()
+    if (normalizedKeyIds.length === 0) {
+      return { models: [], error: '请先选择账号' }
     }
-
-    // 创建新请求
-    const requestPromise = (async (): Promise<FetchResult> => {
-      try {
-        loadingMap.value.set(requestKey, true)
-        const response = await adminApi.queryProviderModels(providerId, apiKeyId, forceRefresh)
-
-        if (response.success && response.data?.models) {
-          const partialWarning = response.data.warning ?? response.data.error
-          return {
-            models: response.data.models,
-            warning: partialWarning ? parseUpstreamModelError(partialWarning) : undefined,
-            fromCache: response.data.from_cache
-          }
-        } else {
-          const rawError = response.data?.error || response.data?.warning || '获取上游模型失败'
-          return { models: [], error: parseUpstreamModelError(rawError) }
-        }
-      } catch (err: unknown) {
-        const rawError = isAxiosError(err) ? (err.response?.data?.detail ?? err.message) : (err instanceof Error ? err.message : String(err))
-        return { models: [], error: parseUpstreamModelError(rawError || '获取上游模型失败') }
-      } finally {
-        loadingMap.value.set(requestKey, false)
-        pendingRequests.delete(requestKey)
-      }
-    })()
-
-    pendingRequests.set(requestKey, requestPromise)
-    return requestPromise
+    const requestKey = getBatchRequestKey(providerId, normalizedKeyIds)
+    return fetchProviderModels(
+      requestKey,
+      forceRefresh,
+      () => adminApi.queryProviderModelsForKeys(
+          providerId,
+          normalizedKeyIds,
+          forceRefresh,
+        ),
+    )
   }
 
   /**
@@ -87,6 +131,7 @@ export function useUpstreamModelsCache() {
 
   return {
     fetchModels,
+    fetchModelsForKeys,
     isLoading,
     loadingMap
   }