fix: keep read refreshes outside write burst

README.md

@@ -1,6 +1,6 @@
 # PicGen Console
 
-一个面向 OpenAI 兼容图像生成 / 编辑接口的本地工作台，当前版本 **0.1.60**。它把
+一个面向 OpenAI 兼容图像生成 / 编辑接口的本地工作台，当前版本 **0.1.61**。它把
 `/v1/images/generations`、`/v1/images/edits` 与 `/v1/responses`（含 `image_generation` 工具）
 包装成统一可观测的代理，前端是一套零依赖的 Web 控制台。
 
@@ -14,7 +14,9 @@
 
 ![PicGen Console 主程序界面](demo1.png)
 
-## 0.1.60 主要特性
+## 0.1.61 主要特性
+
+- **正常页面刷新不再误限流**：业务 API 仍受每分钟总配额保护，5 秒 burst 只约束写请求；生成完成后的统计刷新、工作区重载和作品列表读取不会互相挤占短时写入配额。
 
 - **多候选兼容参数路径**：上游若以 `tools[0].n` 等嵌套参数路径拒绝候选数量，自动去掉 `n` 并并发补齐候选；结构化识别只针对候选数字段，不会误重试内容审核错误。
 
@@ -62,7 +64,7 @@
   - `RequestIdMiddleware` — 生成或透传 `X-Request-ID`
   - `SecurityHeadersMiddleware` — `X-Content-Type-Options / X-Frame-Options / Referrer-Policy / Permissions-Policy / Cache-Control`
   - `BodySizeLimitMiddleware` — 拒绝超大请求（默认 64 MB）
-  - `RateLimitMiddleware` — 滑动窗口限流（默认 120 req/min + 20 burst/5s）
+  - `RateLimitMiddleware` — 滑动窗口限流（默认业务 API 120 req/min + 写请求 20 burst/5s）
   - `ProxyAuthMiddleware` — 可选 Bearer / `X-Proxy-Token` 鉴权
   - `CORSMiddleware` — 配置化跨域
 - **原子化落盘**：图片与 sidecar JSON 用临时文件 + rename 写入，崩溃不留半截文件；核心生成、
@@ -113,10 +115,10 @@ PICGEN_LOG_FORMAT=json \
 ### Docker
 
 ```bash
-docker build -t minorli/picgen:0.1.60 .
+docker build -t minorli/picgen:0.1.61 .
 docker run --rm -p 8000:8000 \
   -v picgen-data:/app/data \
-  minorli/picgen:0.1.60
+  minorli/picgen:0.1.61
 ```
 
 或：
@@ -131,10 +133,10 @@ docker compose up -d
 ./scripts/docker-build-push.sh
 ```
 
-默认会构建并推送 `minorli/picgen:0.1.60`。也可以覆盖：
+默认会构建并推送 `minorli/picgen:0.1.61`。也可以覆盖：
 
 ```bash
-IMAGE=minorli/picgen VERSION=0.1.60 PLATFORM=linux/amd64 ./scripts/docker-build-push.sh
+IMAGE=minorli/picgen VERSION=0.1.61 PLATFORM=linux/amd64 ./scripts/docker-build-push.sh
 ```
 
 镜像不会包含 `.env`、本地用户库或历史图片。容器内置 `HEALTHCHECK` 探测 `/api/health`，以非 root
@@ -247,7 +249,7 @@ Bug 反馈和找回密码申请会先写入本地认证库，再优先发送到
 
 ## 图像通道
 
-PicGen 0.1.60 把四类图像操作统一提交给 `/api/image-jobs`，实际通道由服务端决定：
+PicGen 0.1.61 把四类图像操作统一提交给 `/api/image-jobs`，实际通道由服务端决定：
 
 | 用户操作 | 默认接口 | 默认模型 |
 | --- | --- | --- |

docker-compose.yml

@@ -1,6 +1,6 @@
 services:
   picgen:
-    image: minorli/picgen:0.1.60
+    image: minorli/picgen:0.1.61
     build:
       context: .
     ports:

pyproject.toml

@@ -1,6 +1,6 @@
 [project]
 name = "picgen"
-version = "0.1.60"
+version = "0.1.61"
 description = "Enterprise-grade local web console for OpenAI-compatible image generation and editing APIs."
 readme = "README.md"
 requires-python = ">=3.12"

scripts/docker-build-push.sh

@@ -2,7 +2,7 @@
 set -euo pipefail
 
 IMAGE="${IMAGE:-minorli/picgen}"
-VERSION="${VERSION:-0.1.60}"
+VERSION="${VERSION:-0.1.61}"
 PLATFORM="${PLATFORM:-linux/amd64}"
 
 docker buildx build \

src/picgen/middleware.py

@@ -243,6 +243,7 @@ async def dispatch(self, request: Request, call_next: RequestHandler) -> Respons
         if request.url.path in {"/api/health", "/api/ready", "/api/config"}:
             return await call_next(request)
         client_key = self._client_key(request)
+        uses_burst_limit = request.method.upper() not in {"GET", "HEAD", "OPTIONS"}
         now = time.monotonic()
         async with self._lock:
             if now - self._last_sweep >= 60.0:
@@ -252,7 +253,9 @@ async def dispatch(self, request: Request, call_next: RequestHandler) -> Respons
             burst_bucket = self._burst_buckets.setdefault(client_key, deque())
             self._evict(minute_bucket, now - 60.0)
             self._evict(burst_bucket, now - 5.0)
-            if len(minute_bucket) >= self.per_minute or (self.burst > 0 and len(burst_bucket) >= self.burst):
+            if len(minute_bucket) >= self.per_minute or (
+                uses_burst_limit and self.burst > 0 and len(burst_bucket) >= self.burst
+            ):
                 retry_after = 5
                 if len(minute_bucket) >= self.per_minute and minute_bucket:
                     # The minute window is what's exhausted — tell the client
@@ -275,7 +278,8 @@ async def dispatch(self, request: Request, call_next: RequestHandler) -> Respons
                     extra_headers={"Retry-After": str(retry_after)},
                 )
             minute_bucket.append(now)
-            burst_bucket.append(now)
+            if uses_burst_limit:
+                burst_bucket.append(now)
         return await call_next(request)
 
 

static/app.js

@@ -1,11 +1,11 @@
-import { calculateLogoPlacementScore, chooseLogoPlacement } from "./logo-placement.mjs?v=0.1.60"
+import { calculateLogoPlacementScore, chooseLogoPlacement } from "./logo-placement.mjs?v=0.1.61"
 import {
   DEFAULT_RESPONSES_MODEL,
   RESPONSES_MODEL_STORAGE_VERSION,
   RESPONSES_REASONING_STORAGE_VERSION,
   migrateStoredResponsesReasoningSettings,
   migrateStoredResponsesSettings,
-} from "./responses-settings.mjs?v=0.1.60"
+} from "./responses-settings.mjs?v=0.1.61"
 
 const RESPONSES_REASONING_EFFORTS = new Set(["low", "medium", "high", "xhigh", "max", "ultra"])
 const DEFAULT_RESPONSES_REASONING_EFFORT = "xhigh"

static/index.html

@@ -5,7 +5,7 @@
     <meta name="viewport" content="width=device-width, initial-scale=1">
     <title>PicGen Console</title>
     <link rel="icon" href="data:image/svg+xml,%3Csvg xmlns='http://www.w3.org/2000/svg' viewBox='0 0 64 64'%3E%3Crect width='64' height='64' rx='16' fill='%23108566'/%3E%3Cpath d='M32 13l16 9v20l-16 9-16-9V22l16-9z' fill='%23ecfdf5'/%3E%3Ccircle cx='32' cy='32' r='8' fill='%23108566'/%3E%3C/svg%3E">
-    <link rel="stylesheet" href="styles.css?v=0.1.60">
+    <link rel="stylesheet" href="styles.css?v=0.1.61">
   </head>
   <body class="auth-gate">
     <div id="authOverlay" class="auth-page" aria-hidden="false">
@@ -1046,7 +1046,7 @@ <h2>管理员高级设置</h2>
       </div>
 
       <footer class="system-footer">
-        <span>PicGen Console　v0.1.60</span>
+        <span>PicGen Console　v0.1.61</span>
         <span>本地代理保存结果到 <strong>data/outputs</strong></span>
       </footer>
     </div>
@@ -1356,6 +1356,6 @@ <h2 id="previewModalTitle">图片预览</h2>
       <span id="toastMessageText"></span>
     </div>
 
-    <script src="app.js?v=0.1.60" type="module"></script>
+    <script src="app.js?v=0.1.61" type="module"></script>
   </body>
 </html>

tests/test_api.py

@@ -3392,8 +3392,8 @@ async def send(message: dict[str, object]) -> None:
     assert response_start["status"] == 204
 
 
-def test_rate_limit_returns_429(make_client, settings_factory):
-    settings = settings_factory(rate_limit_per_minute=2, rate_limit_burst=2)
+def test_rate_limit_write_burst_returns_429(make_client, settings_factory):
+    settings = settings_factory(rate_limit_per_minute=100, rate_limit_burst=2)
     client, _, _ = make_client(settings=settings)
     # generate endpoint -> still validation 400, but counts toward rate limit
     client.post("/api/generate", json={})
@@ -3404,6 +3404,44 @@ def test_rate_limit_returns_429(make_client, settings_factory):
     assert "Retry-After" in response.headers
 
 
+def test_rate_limit_burst_does_not_block_read_only_ui_refresh(make_client, settings_factory):
+    settings = settings_factory(rate_limit_per_minute=100, rate_limit_burst=2)
+    client, _, _ = make_client(settings=settings)
+
+    responses = [client.get("/api/recipes") for _ in range(5)]
+
+    assert all(response.status_code != 429 for response in responses)
+
+
+def test_read_requests_do_not_consume_write_burst(make_client, settings_factory):
+    settings = settings_factory(rate_limit_per_minute=100, rate_limit_burst=2)
+    client, _, _ = make_client(settings=settings)
+
+    for _ in range(5):
+        assert client.get("/api/recipes").status_code != 429
+
+    first_write = client.post("/api/generate", json={})
+    second_write = client.post("/api/generate", json={})
+    blocked_write = client.post("/api/generate", json={})
+
+    assert first_write.status_code != 429
+    assert second_write.status_code != 429
+    assert blocked_write.status_code == 429
+
+
+def test_rate_limit_per_minute_still_applies_to_read_requests(make_client, settings_factory):
+    settings = settings_factory(rate_limit_per_minute=2, rate_limit_burst=1)
+    client, _, _ = make_client(settings=settings)
+
+    first = client.get("/api/recipes")
+    second = client.get("/api/recipes")
+    third = client.get("/api/recipes")
+
+    assert first.status_code != 429
+    assert second.status_code != 429
+    assert third.status_code == 429
+
+
 def test_proxy_auth_token_enforced(make_client, settings_factory):
     settings = settings_factory(proxy_auth_token="topsecret")
     client, _, _ = make_client(settings=settings)

tests/test_static_assets.py

@@ -13,7 +13,7 @@ def test_legacy_responses_model_storage_is_migrated_once() -> None:
     settings_js = (ROOT_DIR / "static" / "responses-settings.mjs").read_text(encoding="utf-8")
 
     assert 'const DEPRECATED_RESPONSES_MODELS = new Set(["gpt-5.4"])' in app_js
-    assert 'from "./responses-settings.mjs?v=0.1.60"' in app_js
+    assert 'from "./responses-settings.mjs?v=0.1.61"' in app_js
     assert 'const LEGACY_DEFAULT_RESPONSES_MODEL = "gpt-5.5"' in settings_js
     assert "const RESPONSES_MODEL_STORAGE_VERSION = 4" in settings_js
     assert "function migrateStoredResponsesSettings" in settings_js
@@ -38,7 +38,7 @@ def test_logo_overlay_uses_uploaded_asset_without_ai_guidance() -> None:
     assert 'const COMPANY_LOGO_URL = "6renyou.png"' in app_js
     assert "composeLogoOverlayForCandidates" in app_js
     assert "createOfficialLogoCanvas" in app_js
-    assert 'from "./logo-placement.mjs?v=0.1.60"' in app_js
+    assert 'from "./logo-placement.mjs?v=0.1.61"' in app_js
     assert "chooseLogoPlacement" in app_js
     assert "calculateLogoPlacementScore" in app_js
     assert "expandLogoSafetyRegion" in placement_js