feat(desktop+server): Phase 3.2 Premium LLM — Anthropic Claude 프리미엄 파이프라인 + 모델별 쿼터 + SaaS UI

빅뱅 8/8 마지막 성공 기준 달성. Supabase Edge Function(llm-proxy)을 통해
Anthropic Claude를 호출하는 PremiumLLMService 신규 구현. 사용자가 Settings에서
Local/Premium 백엔드를 선택하면 VoiceModeService가 자동 분기하고, Premium 실패 시
Local로 silent fallback + 상단 중앙 배너 알림.

실측: Claude Haiku refine 1.6~3.2초 (이전 qwen3 42.9초 → 13~27배 빠름).

주요 변경:
- PremiumLLMService 신규 (싱글톤+EventEmitter, processText/chatStream,
  Supabase functions.invoke 기반, _ensureAuth 가드)
- llm-prompts.ts: SYSTEM_PROMPTS를 Local/Premium 공유 모듈로 추출
  (resolveSystemPrompt 헬퍼)
- VoiceModeService: _getLLMProcessor → _runProcessorWithFallback 라우터 +
  premium-llm-fallback 이벤트
- CloudSyncService: getAccessToken(async), getAnonKey, invokeFunction(auth
  헤더 자동 처리, 에러 body 파싱)
- IPC: LLM.PREMIUM_* 채널 6개 + preload API + llm-handlers 이벤트 전달
  (safeSendToRenderer 헬퍼)
- AppConfig.llmBackend: 'local' | 'premium' (기본 'local')
- Settings UI: Backend 드롭다운 + Premium 선택 시 Ollama UI 숨김 + 라이선스
  모달 자동 오픈
- AppLayout: 상단 중앙 Snackbar fallback 배너 (8초, warning filled)
- LicenseModal: 라이선스 키 입력 제거 → SaaS 구독 관리 UI 전환
  (Free/Pro/Pro+ 업그레이드 버튼, Payple 준비 중 스텁)
- 등급 비교 표: featureLabel i18n 번역 수정

서버 (Supabase Edge Functions):
- quota.ts: 모델별 쿼터 구조 (llm_haiku/sonnet/opus × free/pro/pro_plus),
  주간/일간 기간 분리, modelToQuotaKey 매핑, consumeQuota baseLimit 파라미터화
- llm-proxy: 모델별 쿼터 체크 + 소비 (checkQuota → consumeQuota 원자적),
  verify_jwt=false (2026 sb_publishable_ 키 호환)
- config.toml: llm-proxy verify_jwt = false
- migration 20260412000001: tier team→pro_plus 통일, subscriptions.overage_credits
  컬럼, consume_quota RPC (원자적 base→overage fallback)

Tier/쿼터:
- free: Haiku 250/주간, Sonnet/Opus 불가
- pro ₩9,900: Haiku 1500/일, Sonnet 300/일, Opus 50/일
- pro_plus ₩29,900: Haiku 무제한, Sonnet 1500/일, Opus 300/일
- api-client SubscriptionTier: team→pro_plus, overage_credits 필드 추가
This commit is contained in:
윤찬 2026-04-12 18:28:02 +09:00
parent d397bcbf57
commit 6e52c18e5b
23 changed files with 1111 additions and 311 deletions

View file

@ -1,11 +1,19 @@
// server/supabase/functions/llm-proxy/index.ts
// Anthropic Claude Messages API 프록시.
// Phase 3.2: 모델별 쿼터 (Haiku/Sonnet/Opus × Free/Pro/Pro+)
// 요청: application/json { messages, system?, max_tokens?, model? }
// 응답: JSON (non-stream) 또는 SSE (stream=true)
import { corsHeaders, handleCorsPreflightRequest } from '../_shared/cors.ts'
import { requireUser, authErrorResponse, type AuthError } from '../_shared/auth.ts'
import { checkQuota, consumeQuota, createServiceRoleClient, type Tier } from '../_shared/quota.ts'
import {
checkQuota,
consumeQuota,
createServiceRoleClient,
modelToQuotaKey,
getQuotaPolicy,
type Tier,
} from '../_shared/quota.ts'
interface LlmRequest {
messages: Array<{ role: 'user' | 'assistant'; content: string }>
@ -15,17 +23,17 @@ interface LlmRequest {
stream?: boolean
}
/** 티어별 허용 모델 */
/** 티어별 허용 모델 — free는 Haiku만, pro는 +Sonnet, pro_plus는 +Opus */
const TIER_MODELS: Record<Tier, string[]> = {
free: ['claude-haiku-4-5-20251001'],
pro: ['claude-haiku-4-5-20251001', 'claude-sonnet-4-6'],
team: ['claude-haiku-4-5-20251001', 'claude-sonnet-4-6', 'claude-opus-4-6']
pro: ['claude-haiku-4-5-20251001', 'claude-sonnet-4-6', 'claude-opus-4-6'],
pro_plus: ['claude-haiku-4-5-20251001', 'claude-sonnet-4-6', 'claude-opus-4-6'],
}
const DEFAULT_MODEL: Record<Tier, string> = {
free: 'claude-haiku-4-5-20251001',
pro: 'claude-sonnet-4-6',
team: 'claude-sonnet-4-6'
pro_plus: 'claude-sonnet-4-6',
}
// @ts-expect-error — Deno 런타임 전역
@ -36,45 +44,75 @@ Deno.serve(async (req: Request) => {
if (req.method !== 'POST') {
return new Response(JSON.stringify({ error: 'Method not allowed' }), {
status: 405,
headers: { ...corsHeaders, 'Content-Type': 'application/json' }
headers: { ...corsHeaders, 'Content-Type': 'application/json' },
})
}
try {
const user = await requireUser(req)
const serviceClient = createServiceRoleClient()
const quota = await checkQuota(user.id, 'llm_process', serviceClient)
if (!quota.allowed) {
return new Response(
JSON.stringify({
error: 'quota_exceeded',
current: quota.current,
limit: quota.limit,
tier: quota.tier
}),
{
status: 429,
headers: { ...corsHeaders, 'Content-Type': 'application/json' }
}
)
}
const body = (await req.json()) as LlmRequest
// 1단계: 티어 조회 (어떤 모델이든 한 번만 읽으면 됨 — haiku로 대리 조회)
const tierCheck = await checkQuota(user.id, 'llm_haiku', serviceClient)
const tier = tierCheck.tier
// 모델 선택 + 티어 검증
const requestedModel = body.model ?? DEFAULT_MODEL[quota.tier]
if (!TIER_MODELS[quota.tier].includes(requestedModel)) {
const requestedModel = body.model ?? DEFAULT_MODEL[tier]
if (!TIER_MODELS[tier].includes(requestedModel)) {
return new Response(
JSON.stringify({
error: 'model_not_allowed',
tier: quota.tier,
tier,
requested: requestedModel,
allowed: TIER_MODELS[quota.tier]
allowed: TIER_MODELS[tier],
}),
{
status: 403,
headers: { ...corsHeaders, 'Content-Type': 'application/json' }
}
headers: { ...corsHeaders, 'Content-Type': 'application/json' },
},
)
}
// 2단계: 해당 모델의 쿼터 확인 (모델별 일간/주간)
const quotaKey = modelToQuotaKey(requestedModel)
const modelQuota = await checkQuota(user.id, quotaKey, serviceClient)
if (!modelQuota.allowed) {
return new Response(
JSON.stringify({
error: 'quota_exceeded',
model: requestedModel,
current: modelQuota.current,
limit: modelQuota.limit,
period: modelQuota.period,
tier,
overage_credits: modelQuota.overageCredits,
}),
{
status: 429,
headers: { ...corsHeaders, 'Content-Type': 'application/json' },
},
)
}
// 3단계: 쿼터 소비 (원자적 base → overage fallback)
const policy = getQuotaPolicy(tier, quotaKey)
const consume = await consumeQuota(user.id, quotaKey, serviceClient, policy.limit)
if (!consume.allowed) {
return new Response(
JSON.stringify({
error: 'quota_exceeded',
model: requestedModel,
current: consume.current,
limit: consume.limit,
tier,
overage_credits: consume.overageCredits,
}),
{
status: 429,
headers: { ...corsHeaders, 'Content-Type': 'application/json' },
},
)
}
@ -82,12 +120,9 @@ Deno.serve(async (req: Request) => {
// @ts-expect-error — Deno.env
const anthropicKey = Deno.env.get('ANTHROPIC_API_KEY') ?? ''
await consumeQuota(user.id, 'llm_process', serviceClient, 1)
if (!anthropicKey) {
// Placeholder 응답 (키 미설정 시)
if (body.stream) {
// 스트리밍 placeholder — SSE로 "설정되지 않음" 메시지 전송
const encoder = new TextEncoder()
const stream = new ReadableStream({
start(controller) {
@ -95,13 +130,13 @@ Deno.serve(async (req: Request) => {
for (const ch of msg) {
controller.enqueue(
encoder.encode(
`data: ${JSON.stringify({ type: 'content_block_delta', delta: { type: 'text_delta', text: ch } })}\n\n`
)
`data: ${JSON.stringify({ type: 'content_block_delta', delta: { type: 'text_delta', text: ch } })}\n\n`,
),
)
}
controller.enqueue(encoder.encode('data: [DONE]\n\n'))
controller.close()
}
},
})
return new Response(stream, {
status: 200,
@ -109,8 +144,8 @@ Deno.serve(async (req: Request) => {
...corsHeaders,
'Content-Type': 'text/event-stream',
'Cache-Control': 'no-cache',
Connection: 'keep-alive'
}
Connection: 'keep-alive',
},
})
}
@ -122,16 +157,16 @@ Deno.serve(async (req: Request) => {
content: [
{
type: 'text',
text: '[llm-proxy placeholder — ANTHROPIC_API_KEY 미설정]'
}
text: '[llm-proxy placeholder — ANTHROPIC_API_KEY 미설정]',
},
],
stop_reason: 'end_turn',
usage: { input_tokens: 0, output_tokens: 0 }
usage: { input_tokens: 0, output_tokens: 0 },
}),
{
status: 200,
headers: { ...corsHeaders, 'Content-Type': 'application/json' }
}
headers: { ...corsHeaders, 'Content-Type': 'application/json' },
},
)
}
@ -141,15 +176,15 @@ Deno.serve(async (req: Request) => {
headers: {
'Content-Type': 'application/json',
'x-api-key': anthropicKey,
'anthropic-version': '2023-06-01'
'anthropic-version': '2023-06-01',
},
body: JSON.stringify({
model: requestedModel,
max_tokens: body.max_tokens ?? 2048,
system: body.system,
messages: body.messages,
stream: body.stream ?? false
})
stream: body.stream ?? false,
}),
})
if (!anthropicResp.ok) {
@ -158,22 +193,21 @@ Deno.serve(async (req: Request) => {
}
if (body.stream && anthropicResp.body) {
// SSE 스트림을 그대로 전달
return new Response(anthropicResp.body, {
status: 200,
headers: {
...corsHeaders,
'Content-Type': 'text/event-stream',
'Cache-Control': 'no-cache',
Connection: 'keep-alive'
}
Connection: 'keep-alive',
},
})
}
const data = await anthropicResp.json()
return new Response(JSON.stringify(data), {
status: 200,
headers: { ...corsHeaders, 'Content-Type': 'application/json' }
headers: { ...corsHeaders, 'Content-Type': 'application/json' },
})
} catch (err) {
if (err && typeof err === 'object' && 'status' in err && 'message' in err) {
@ -182,7 +216,7 @@ Deno.serve(async (req: Request) => {
const message = err instanceof Error ? err.message : 'Unknown error'
return new Response(JSON.stringify({ error: message }), {
status: 500,
headers: { ...corsHeaders, 'Content-Type': 'application/json' }
headers: { ...corsHeaders, 'Content-Type': 'application/json' },
})
}
})