fix channel quota and ark inference

apps/electron/src/main/lib/channel-manager.test.ts

@@ -0,0 +1,129 @@
+import { afterAll, beforeAll, beforeEach, describe, expect, mock, test } from 'bun:test'
+import { mkdirSync, mkdtempSync, rmSync, writeFileSync } from 'node:fs'
+import * as os from 'node:os'
+import { join } from 'node:path'
+
+type ChannelManagerModule = typeof import('./channel-manager')
+
+let channelManager: ChannelManagerModule
+let tempHome: string
+let lastFetchInput: { url: string; init?: RequestInit } | null = null
+const originalHome = process.env.HOME
+const originalPromaDev = process.env.PROMA_DEV
+
+mock.module('electron', () => ({
+  safeStorage: {
+    isEncryptionAvailable: () => false,
+    encryptString: (value: string) => Buffer.from(value),
+    decryptString: (value: Buffer) => value.toString('utf-8'),
+  },
+}))
+
+mock.module('node:os', () => ({
+  ...os,
+  homedir: () => tempHome,
+}))
+
+mock.module('./proxy-settings-service', () => ({
+  getEffectiveProxyUrl: async () => undefined,
+}))
+
+mock.module('./proxy-fetch', () => ({
+  getFetchFn: () => async (url: string, init?: RequestInit) => {
+    lastFetchInput = { url, init }
+    return new Response(JSON.stringify({
+      code: 200,
+      success: true,
+      data: {
+        limits: [
+          {
+            type: 'TOKENS_LIMIT',
+            unit: 3,
+            percentage: 25,
+            nextResetTime: 1_800_000_000,
+          },
+          {
+            type: 'TOKENS_LIMIT',
+            unit: 6,
+            percentage: 40,
+            nextResetTime: 1_800_086_400,
+          },
+        ],
+      },
+    }), {
+      status: 200,
+      headers: { 'Content-Type': 'application/json' },
+    })
+  },
+}))
+
+beforeAll(async () => {
+  tempHome = mkdtempSync(join(os.tmpdir(), 'proma-channel-manager-'))
+  process.env.HOME = tempHome
+  process.env.PROMA_DEV = '0'
+  channelManager = await import('./channel-manager')
+})
+
+beforeEach(() => {
+  lastFetchInput = null
+  rmSync(join(tempHome, '.proma'), { recursive: true, force: true })
+  mkdirSync(join(tempHome, '.proma'), { recursive: true })
+})
+
+afterAll(() => {
+  if (originalHome === undefined) {
+    delete process.env.HOME
+  } else {
+    process.env.HOME = originalHome
+  }
+  if (originalPromaDev === undefined) {
+    delete process.env.PROMA_DEV
+  } else {
+    process.env.PROMA_DEV = originalPromaDev
+  }
+  rmSync(tempHome, { recursive: true, force: true })
+})
+
+describe('Channel manager plan quota', () => {
+  test('Given 火山方舟 Anthropic 兼容地址 When 拉取模型 Then 使用 Ark Coding Plan 预设', async () => {
+    const result = await channelManager.fetchModels({
+      provider: 'anthropic-compatible',
+      baseUrl: 'https://ark.cn-beijing.volces.com/api/coding',
+      apiKey: 'ark-token',
+    })
+
+    expect(result.success).toBe(true)
+    expect(result.message).toContain('火山方舟 Coding Plan')
+    expect(result.models.map((model) => model.id)).toContain('doubao-seed-2.0-code')
+    expect(lastFetchInput).toBeNull()
+  })
+
+  test('Given 智谱团队版只填写 API Token When 查询额度 Then 使用默认组织项目上下文', async () => {
+    writeFileSync(join(tempHome, '.proma', 'channels.json'), JSON.stringify({
+      version: 2,
+      channels: [
+        {
+          id: 'zhipu-team',
+          name: 'GLM Team',
+          provider: 'zhipu-coding-team',
+          baseUrl: 'https://open.bigmodel.cn/api/anthropic',
+          apiKey: 'team-token',
+          models: [],
+          enabled: true,
+          createdAt: 1,
+          updatedAt: 1,
+        },
+      ],
+    }), 'utf-8')
+
+    const result = await channelManager.getChannelPlanQuota('zhipu-team')
+
+    expect(result.supported).toBe(true)
+    expect(result.provider).toBe('zhipu-coding-team')
+    expect(result.windows.map((window) => window.type)).toEqual(['5h', 'weekly'])
+    expect(lastFetchInput?.url).toBe('https://bigmodel.cn/api/monitor/usage/quota/limit?type=2')
+    expect((lastFetchInput?.init?.headers as Record<string, string>).Authorization).toBe('team-token')
+    expect((lastFetchInput?.init?.headers as Record<string, string>)['bigmodel-organization']).toBeUndefined()
+    expect((lastFetchInput?.init?.headers as Record<string, string>)['bigmodel-project']).toBeUndefined()
+  })
+})

apps/electron/src/main/lib/channel-manager.ts

@@ -131,6 +131,11 @@ function inferProviderFromBaseUrl(provider: ProviderType, baseUrl: string): Prov
     if (hostname === 'api.moonshot.cn' || hostname === 'api.moonshot.ai') {
       return 'kimi-api'
     }
+    // 火山方舟域名同时承载 OpenAI 协议与 Anthropic Coding Plan。只有用户选了
+    // 通用 Anthropic 兼容格式时才纠正为内置 ark-coding-plan，避免拉模型误走 /models。
+    if (provider === 'anthropic-compatible' && hostname === 'ark.cn-beijing.volces.com') {
+      return 'ark-coding-plan'
+    }
     return provider
   } catch {
     return provider
@@ -1201,8 +1206,10 @@ async function queryZhipuPlanQuota(
   provider: ProviderType = 'zhipu-coding',
 ): Promise<ChannelPlanQuotaResult> {
   if (provider === 'zhipu-coding-team') {
-    const teamCredentials = parseZhipuTeamCredentials(apiKey)
-    if (!teamCredentials) {
+    const teamCredentials = parseZhipuTeamCredentials(apiKey) ?? {
+      apiKey: extractZhipuCodingTeamApiToken(apiKey),
+    }
+    if (!teamCredentials.apiKey.trim()) {
       return createUnsupportedPlanQuota(
         'zhipu-coding-team',
         '请填写 API Token；组织 ID 和项目 ID 可选',