u1s1-cli 0.13.6 → 0.13.7

package/dist/agent-setup.d.ts

@@ -21,7 +21,10 @@ export declare function cleanupBrandThemes(): void;
  * 环境变量,文件里不落密钥。U1S1_TOOLS_VIA_EXTENSION 守卫:旧版 CLI 仍在
  * 进程内注册工具且不设该变量,残留的本文件在旧版下自动空转,避免双重注册。
  */
-export declare function writeWebToolsExtension(cfg: CliConfig, webSearchEnabled: boolean): void;
+export declare function writeWebToolsExtension(cfg: CliConfig, features: {
+    webSearch: boolean;
+    webFetchRender: boolean;
+}): void;
 /**
  * pi-web-ui 的就绪检测只认 <agentDir>/auth.json 里有没有凭据条目(models.json
  * 的 apiKey 它不看),空 {} 会在网页里弹「未检测到 pi agent 配置」引导装 pi。

package/dist/agent-setup.js

@@ -123,11 +123,11 @@ export function cleanupBrandThemes() {
  * 环境变量,文件里不落密钥。U1S1_TOOLS_VIA_EXTENSION 守卫:旧版 CLI 仍在
  * 进程内注册工具且不设该变量,残留的本文件在旧版下自动空转,避免双重注册。
  */
-export function writeWebToolsExtension(cfg, webSearchEnabled) {
+export function writeWebToolsExtension(cfg, features) {
     const dir = join(agentDir, "extensions");
     mkdirSync(dir, { recursive: true });
     const toolsUrl = new URL("./tools.js", import.meta.url).href;
-    const searchLine = webSearchEnabled
+    const searchLine = features.webSearch
         ? `  pi.registerTool(tools.createSearchTool({ baseUrl: ${JSON.stringify(cfg.baseUrl)}, apiKey: process.env.U1S1_API_KEY }));\n`
         : "";
     writeFileSync(join(dir, "u1s1-tools.js"), `// 由 u1s1 每次启动自动生成,请勿手改\n` +
@@ -135,7 +135,7 @@ export function writeWebToolsExtension(cfg, webSearchEnabled) {
         `  if (process.env.U1S1_TOOLS_VIA_EXTENSION !== "1") return;\n` +
         `  const tools = await import(${JSON.stringify(toolsUrl)});\n` +
         searchLine +
-        `  pi.registerTool(tools.webFetchTool);\n` +
+        `  pi.registerTool(tools.createFetchTool({ baseUrl: ${JSON.stringify(cfg.baseUrl)}, apiKey: process.env.U1S1_API_KEY, renderFallback: ${features.webFetchRender} }));\n` +
         `}\n`);
 }
 /**

package/dist/api.d.ts

@@ -46,6 +46,8 @@ export interface ApiModel {
 /** 服务端能力开关;字段缺失(老网关)时按开启处理,保持现状。 */
 export interface ApiFeatures {
     web_search?: boolean;
+    /** 网关配了 Browser Rendering 才为 true;老网关没有 /v1/fetch,缺失按关闭处理。 */
+    web_fetch_render?: boolean;
 }
 export interface ModelsResponse {
     models: ApiModel[];
@@ -88,4 +90,9 @@ export interface SearchResponse {
 }
 /** 联网搜索走网关代理（上游 key 只在服务端）。maxResults 不传时由服务端决定默认值。 */
 export declare function searchWeb(cfg: Pick<CliConfig, "baseUrl" | "apiKey">, query: string, maxResults?: number, signal?: AbortSignal): Promise<SearchResponse>;
+/** web_fetch 直连失败时的回退:网关用 Cloudflare Browser Rendering 渲染后转 markdown。 */
+export declare function renderPage(cfg: Pick<CliConfig, "baseUrl" | "apiKey">, url: string, signal?: AbortSignal): Promise<{
+    url: string;
+    markdown: string;
+}>;
 export declare function fetchMe(cfg: CliConfig): Promise<MeResponse>;

package/dist/api.js

@@ -80,6 +80,33 @@ export async function searchWeb(cfg, query, maxResults, signal) {
     }
     return (await resp.json());
 }
+/** web_fetch 直连失败时的回退:网关用 Cloudflare Browser Rendering 渲染后转 markdown。 */
+export async function renderPage(cfg, url, signal) {
+    if (!cfg.apiKey)
+        throw new Error("没有配置 API Key");
+    let resp;
+    try {
+        resp = await fetch(`${cfg.baseUrl}/fetch`, {
+            method: "POST",
+            headers: {
+                ...authHeaders(cfg.apiKey),
+                "content-type": "application/json",
+            },
+            body: JSON.stringify({ url }),
+            signal,
+        });
+    }
+    catch {
+        throw new Error(`连不上 ${cfg.baseUrl}`);
+    }
+    if (resp.status === 401)
+        throw new Error("这把 Key 不对或已失效,去 https://u1s1.io/dashboard 看看");
+    if (!resp.ok) {
+        const body = (await resp.json().catch(() => null));
+        throw new Error(body?.error?.message ?? `渲染服务返回 ${resp.status}`);
+    }
+    return (await resp.json());
+}
 export async function fetchMe(cfg) {
     if (!cfg.apiKey)
         throw new Error("没有配置 API Key");

package/dist/index.js

@@ -145,11 +145,14 @@ async function runAgent(cfg, args) {
     // 服务端没开搜索(或老网关没有 features 字段)时不注册 web_search,模型就不会白调
     // 自定义端点与模型列表并行拉取,失败各自兜底(内置列表 / 本地缓存)
     let webSearchEnabled = true;
+    // 老网关没有 /v1/fetch,web_fetch_render 缺失时按关闭处理,直连失败不去白调
+    let webFetchRenderEnabled = false;
     const endpointsReady = loadCustomEndpoints(cfg);
     try {
         const { models, features } = await fetchModels(cfg);
         setModelsFromApi(models.map(apiModelToDef));
         webSearchEnabled = features.web_search !== false;
+        webFetchRenderEnabled = features.web_fetch_render === true;
     }
     catch (e) {
         console.error("  获取模型列表失败,使用内置列表:", e.message);
@@ -158,7 +161,7 @@ async function runAgent(cfg, args) {
     ensureBrandPrompt(await shellReady);
     ensureProviderModels(cfg);
     // 联网工具经 agentDir/extensions 投影,TUI 和 u1s1 web 共用一份注册
-    writeWebToolsExtension(cfg, webSearchEnabled);
+    writeWebToolsExtension(cfg, { webSearch: webSearchEnabled, webFetchRender: webFetchRenderEnabled });
     ensureTmuxKeyboardProtocol();
     // must be set before pi reads them (getAgentDir() reads at call time, env at import is fine too)
     process.env["PI_CODING_AGENT_DIR"] = agentDir;

package/dist/tools.d.ts

@@ -8,11 +8,21 @@ export declare function createSearchTool(cfg: Pick<CliConfig, "baseUrl" | "apiKe
     query: string;
     count: number;
 }, any> & import("@earendil-works/pi-coding-agent").ToolDefinition<any, any, any>;
-/** 抓网页:纯客户端出网,不经过我们的服务器,也不额外计费。 */
-export declare const webFetchTool: import("@earendil-works/pi-coding-agent").ToolDefinition<Type.TObject<{
+export interface FetchToolConfig {
+    baseUrl: string;
+    apiKey?: string;
+    /** 网关配了 Browser Rendering(features.web_fetch_render)才开;老网关没有 /v1/fetch。 */
+    renderFallback?: boolean;
+}
+/**
+ * 抓网页工具:先直连(免费、快),打不开或疑似被反爬/JS 空壳时,回退到
+ * 网关的 Cloudflare Browser Rendering 渲染(真无头浏览器,按次计费)。
+ */
+export declare function createFetchTool(cfg: FetchToolConfig): import("@earendil-works/pi-coding-agent").ToolDefinition<Type.TObject<{
     url: Type.TString;
 }>, {
     url: string;
     contentType: string;
+    via: "direct" | "render";
     chars: number;
 }, any> & import("@earendil-works/pi-coding-agent").ToolDefinition<any, any, any>;

package/dist/tools.js

@@ -1,9 +1,13 @@
 import { defineTool } from "@earendil-works/pi-coding-agent";
 import { Type } from "typebox";
-import { searchWeb } from "./api.js";
+import { renderPage, searchWeb } from "./api.js";
 const FETCH_TIMEOUT_MS = 20_000;
 const MAX_FETCH_BYTES = 2_000_000;
 const MAX_TEXT_CHARS = 30_000;
+/** 云端渲染是真开浏览器,比直连慢得多,给宽裕些。 */
+const RENDER_TIMEOUT_MS = 45_000;
+/** 200 但榨出的正文比这还短,多半是 JS 渲染的空壳页,值得上浏览器再试。 */
+const MIN_HTML_TEXT_CHARS = 200;
 const HTML_ENTITIES = {
     " ": " ",
     "&": "&",
@@ -104,59 +108,113 @@ export function createSearchTool(cfg) {
         },
     });
 }
-/** 抓网页:纯客户端出网,不经过我们的服务器,也不额外计费。 */
-export const webFetchTool = defineTool({
-    name: "web_fetch",
-    label: "读取网页",
-    description: "Fetch a URL and return its readable text content (HTML is stripped to text; JSON and plain text are returned as-is). Use it to read a page found via web_search, or any URL the user pasted.",
-    promptSnippet: "抓取指定网址并转成可读文本",
-    promptGuidelines: [
-        "Use web_fetch to read a specific URL, and web_search when you still need to find the URL.",
-    ],
-    parameters: Type.Object({
-        url: Type.String({ description: "Absolute http(s) URL to fetch." }),
-    }),
-    async execute(_toolCallId, params, signal) {
-        let url;
-        try {
-            // 用户从聊天工具粘来的链接常带 "@https://…" 前缀,顺手剥掉
-            url = new URL(params.url.trim().replace(/^@/, ""));
-        }
-        catch {
-            throw new Error(`不是合法的网址: ${params.url}`);
-        }
-        if (url.protocol !== "http:" && url.protocol !== "https:") {
-            throw new Error(`只支持 http/https,收到 ${url.protocol}`);
-        }
-        const timeout = AbortSignal.timeout(FETCH_TIMEOUT_MS);
+/** 直连失败里值得换浏览器再试的一类:网络不通/超时,或典型的反爬状态码。 */
+class DirectFetchError extends Error {
+    renderWorthy;
+    constructor(message, renderWorthy) {
+        super(message);
+        this.renderWorthy = renderWorthy;
+    }
+}
+/** 反爬/风控常用的状态码;404/500 这类真错误换浏览器也救不回来,不白花一次渲染。 */
+const RENDER_WORTHY_STATUS = new Set([403, 405, 406, 429, 503]);
+/** 直连抓取:纯客户端出网,不经过我们的服务器,也不额外计费。 */
+async function fetchDirect(url, signal) {
+    const timeout = AbortSignal.timeout(FETCH_TIMEOUT_MS);
+    const abort = signal ? AbortSignal.any([signal, timeout]) : timeout;
+    let resp;
+    try {
+        resp = await fetch(url, {
+            redirect: "follow",
+            headers: {
+                accept: "text/html,application/xhtml+xml,application/json,text/plain;q=0.9,*/*;q=0.8",
+                "user-agent": "u1s1-cli",
+            },
+            signal: abort,
+        });
+    }
+    catch (e) {
+        // 用户主动打断不算失败,直接往外抛,不去碰云端渲染
+        if (signal?.aborted)
+            throw e;
+        throw new DirectFetchError(`打不开 ${url.href}: ${e.message}`, true);
+    }
+    if (!resp.ok) {
+        throw new DirectFetchError(`${url.href} 返回 ${resp.status} ${resp.statusText}`, RENDER_WORTHY_STATUS.has(resp.status));
+    }
+    const type = resp.headers.get("content-type") ?? "";
+    if (!/text\/|json|xml|javascript/i.test(type)) {
+        throw new DirectFetchError(`${url.href} 不是文本内容 (${type || "unknown"}),读不了`, false);
+    }
+    const raw = decodeBody(await readBodyCapped(resp, MAX_FETCH_BYTES), type);
+    const isHtml = /html|xml/i.test(type);
+    return { text: isHtml ? htmlToText(raw) : raw.trim(), contentType: type, isHtml };
+}
+/**
+ * 抓网页工具:先直连(免费、快),打不开或疑似被反爬/JS 空壳时,回退到
+ * 网关的 Cloudflare Browser Rendering 渲染(真无头浏览器,按次计费)。
+ */
+export function createFetchTool(cfg) {
+    const canRender = () => !!(cfg.renderFallback && cfg.apiKey);
+    async function renderViaGateway(url, signal) {
+        const timeout = AbortSignal.timeout(RENDER_TIMEOUT_MS);
         const abort = signal ? AbortSignal.any([signal, timeout]) : timeout;
-        let resp;
-        try {
-            resp = await fetch(url, {
-                redirect: "follow",
-                headers: {
-                    accept: "text/html,application/xhtml+xml,application/json,text/plain;q=0.9,*/*;q=0.8",
-                    "user-agent": "u1s1-cli",
-                },
-                signal: abort,
+        const { markdown } = await renderPage(cfg, url.href, abort);
+        return markdown.trim();
+    }
+    return defineTool({
+        name: "web_fetch",
+        label: "读取网页",
+        description: "Fetch a URL and return its readable text content (HTML is stripped to text; JSON and plain text are returned as-is). Use it to read a page found via web_search, or any URL the user pasted. If the direct fetch fails or the page needs JavaScript, it automatically retries through a cloud headless browser.",
+        promptSnippet: "抓取指定网址并转成可读文本",
+        promptGuidelines: [
+            "Use web_fetch to read a specific URL, and web_search when you still need to find the URL.",
+        ],
+        parameters: Type.Object({
+            url: Type.String({ description: "Absolute http(s) URL to fetch." }),
+        }),
+        async execute(_toolCallId, params, signal) {
+            let url;
+            try {
+                // 用户从聊天工具粘来的链接常带 "@https://…" 前缀,顺手剥掉
+                url = new URL(params.url.trim().replace(/^@/, ""));
+            }
+            catch {
+                throw new Error(`不是合法的网址: ${params.url}`);
+            }
+            if (url.protocol !== "http:" && url.protocol !== "https:") {
+                throw new Error(`只支持 http/https,收到 ${url.protocol}`);
+            }
+            const ok = (text, via, contentType) => ({
+                content: [{ type: "text", text: truncate(`# ${url.href}\n\n${text || "(空白页面)"}`) }],
+                details: { url: url.href, contentType, via, chars: text.length },
             });
-        }
-        catch (e) {
-            throw new Error(`打不开 ${url.href}: ${e.message}`);
-        }
-        if (!resp.ok)
-            throw new Error(`${url.href} 返回 ${resp.status} ${resp.statusText}`);
-        const type = resp.headers.get("content-type") ?? "";
-        if (!/text\/|json|xml|javascript/i.test(type)) {
-            throw new Error(`${url.href} 不是文本内容 (${type || "unknown"}),读不了`);
-        }
-        const raw = decodeBody(await readBodyCapped(resp, MAX_FETCH_BYTES), type);
-        const text = /html|xml/i.test(type) ? htmlToText(raw) : raw.trim();
-        return {
-            content: [
-                { type: "text", text: truncate(`# ${url.href}\n\n${text || "(空白页面)"}`) },
-            ],
-            details: { url: url.href, contentType: type, chars: text.length },
-        };
-    },
-});
+            let direct = null;
+            try {
+                direct = await fetchDirect(url, signal);
+            }
+            catch (e) {
+                if (!(e instanceof DirectFetchError) || !e.renderWorthy || !canRender())
+                    throw e;
+                try {
+                    return ok(await renderViaGateway(url, signal), "render", "text/markdown");
+                }
+                catch (re) {
+                    throw new Error(`${e.message}；云端浏览器渲染也失败: ${re.message}`);
+                }
+            }
+            // 直连 200 但正文近乎空:大概率是 JS 渲染的 SPA,换浏览器再试;渲染失败就退回空壳
+            if (direct.isHtml && direct.text.length < MIN_HTML_TEXT_CHARS && canRender()) {
+                try {
+                    const rendered = await renderViaGateway(url, signal);
+                    if (rendered.length > direct.text.length)
+                        return ok(rendered, "render", "text/markdown");
+                }
+                catch {
+                    // 直连结果虽薄但还在,渲染挂了不至于让整次调用失败
+                }
+            }
+            return ok(direct.text, "direct", direct.contentType);
+        },
+    });
+}

package/dist/web.js

@@ -30,11 +30,14 @@ export async function prepareWebEnv(cfg) {
     // 服务端没开搜索(或老网关没有 features 字段)时不注册 web_search
     // 自定义端点与模型列表并行拉取,失败各自兜底(内置列表 / 本地缓存)
     let webSearchEnabled = true;
+    // 老网关没有 /v1/fetch,web_fetch_render 缺失时按关闭处理,直连失败不去白调
+    let webFetchRenderEnabled = false;
     const endpointsReady = loadCustomEndpoints(cfg);
     try {
         const { models, features } = await fetchModels(cfg);
         setModelsFromApi(models.map(apiModelToDef));
         webSearchEnabled = features.web_search !== false;
+        webFetchRenderEnabled = features.web_fetch_render === true;
     }
     catch (e) {
         console.error("  获取模型列表失败,使用内置列表:", e.message);
@@ -45,7 +48,7 @@ export async function prepareWebEnv(cfg) {
     // pi-web-ui 靠 auth.json 判断「已配置」,否则网页会弹 pi 安装引导
     ensureAuthCredential();
     // 联网工具经 agentDir/extensions 投影,和 TUI 共用一份注册
-    writeWebToolsExtension(cfg, webSearchEnabled);
+    writeWebToolsExtension(cfg, { webSearch: webSearchEnabled, webFetchRender: webFetchRenderEnabled });
     // 网页版新会话从 settings.json 的 defaultModel 取模型(TUI 是每次传 --model),
     // 确保它有值;resolvePreferredModel 优先尊重已有的 in-session 选择,不会回退覆盖。
     const pref = resolvePreferredModel(cfg);

package/package.json

@@ -1,6 +1,6 @@
 {
   "name": "u1s1-cli",
-  "version": "0.13.6",
+  "version": "0.13.7",
   "description": "u1s1 — 有一说一,最省心的 AI 编程搭子。终端里用中文说需求,AI 帮你读文件、改代码、跑命令。",
   "type": "module",
   "bin": {