nolo-cli 0.1.40 → 0.1.42

package/ai/tools/cfBrowserTools.ts

@@ -0,0 +1,319 @@
+// packages/ai/tools/cfBrowserTools.ts
+// Cloudflare Browser Rendering 工具：截图、Markdown、PDF、JSON 结构化提取
+// Docs: https://developers.cloudflare.com/browser-rendering/rest-api/
+
+import { callToolApi } from "./toolApiClient";
+import {
+  assertFetchableDocsUrl,
+  detectExtractionIssue,
+  discoverCanonicalDocsUrl,
+} from "./fetchWebpageSupport";
+
+async function resolveCfBrowserUrl(url?: string) {
+  if (!url) {
+    return {
+      targetUrl: url,
+      originalUrl: url,
+      rewritten: false,
+    };
+  }
+
+  const resolution = await discoverCanonicalDocsUrl(url);
+  const targetUrl = resolution.resolvedUrl;
+  await assertFetchableDocsUrl(targetUrl, fetch, url);
+
+  return {
+    targetUrl,
+    originalUrl: url,
+    rewritten: targetUrl !== url,
+  };
+}
+
+function buildNormalizationNote(originalUrl?: string, targetUrl?: string) {
+  if (!originalUrl || !targetUrl || originalUrl === targetUrl) return "";
+  return `\n🧭 文档地址已规范化: ${originalUrl} → ${targetUrl}`;
+}
+
+// ──────────────────────────────────────────────────────────
+// cfScreenshot
+// ──────────────────────────────────────────────────────────
+export const cfScreenshotFunctionSchema = {
+  name: "cfScreenshot",
+  description:
+    "使用 Cloudflare Browser Rendering 对指定网页或 HTML 内容截图，支持全页截图、自定义视口。" +
+    "返回 base64 图片数据，可直接在对话中展示。",
+  parameters: {
+    type: "object",
+    properties: {
+      url: {
+        type: "string",
+        description: "要截图的网页 URL（http/https），与 html 二选一。",
+      },
+      html: {
+        type: "string",
+        description: "直接传入 HTML 字符串进行渲染截图，与 url 二选一。",
+      },
+      fullPage: {
+        type: "boolean",
+        description: "是否截取完整页面（包括滚动区域），默认 false 只截可视区。",
+        default: false,
+      },
+      viewport: {
+        type: "object",
+        description: "浏览器视口尺寸，例如 { width: 1280, height: 720 }。",
+        properties: {
+          width: { type: "integer" },
+          height: { type: "integer" },
+        },
+      },
+    },
+  },
+};
+
+export async function cfScreenshotFunc(
+  args: {
+    url?: string;
+    html?: string;
+    fullPage?: boolean;
+    viewport?: { width: number; height: number };
+  },
+  thunkApi: any
+): Promise<{ rawData: any; displayData: string }> {
+  const { url, html, fullPage = false, viewport } = args;
+
+  if (!url && !html) throw new Error("必须提供 url 或 html 参数");
+  const { targetUrl, originalUrl, rewritten } = await resolveCfBrowserUrl(url);
+
+  const data = await callToolApi<{
+    dataUrl: string;
+    mimeType: string;
+    browserMsUsed?: number;
+    source: string;
+  }>(
+    thunkApi,
+    "/api/cf-screenshot",
+    { url: targetUrl, html, fullPage, viewport },
+    { withAuth: true }
+  );
+
+  const seconds = data.browserMsUsed ? (data.browserMsUsed / 1000).toFixed(2) : "?";
+
+  return {
+    rawData: rewritten ? { ...data, originalUrl, resolvedUrl: targetUrl } : data,
+    displayData:
+      `📸 截图完成\n- 来源: ${data.source}\n- 浏览器耗时: ${seconds}s` +
+      buildNormalizationNote(originalUrl, targetUrl) +
+      `\n\n` +
+      `![screenshot](${data.dataUrl})`,
+  };
+}
+
+// ──────────────────────────────────────────────────────────
+// cfGetMarkdown
+// ──────────────────────────────────────────────────────────
+export const cfGetMarkdownFunctionSchema = {
+  name: "cfGetMarkdown",
+  description:
+    "使用 Cloudflare Browser Rendering 将网页内容转换为 Markdown 格式，支持 JS 渲染。" +
+    "比 cloudflareCrawl 更快（单页），比 fetchWebpage 更准确（支持 JS 动态内容）。" +
+    "适合文章阅读、内容总结、单页数据提取。",
+  parameters: {
+    type: "object",
+    properties: {
+      url: {
+        type: "string",
+        description: "要读取的网页 URL（http/https），与 html 二选一。",
+      },
+      html: {
+        type: "string",
+        description: "直接传入 HTML 字符串转换，与 url 二选一。",
+      },
+      waitForNetworkIdle: {
+        type: "boolean",
+        description: "是否等待网络请求结束再提取（适合 SPA/动态页面），默认 false。",
+        default: false,
+      },
+    },
+  },
+};
+
+export async function cfGetMarkdownFunc(
+  args: {
+    url?: string;
+    html?: string;
+    waitForNetworkIdle?: boolean;
+  },
+  thunkApi: any
+): Promise<{ rawData: any; displayData: string }> {
+  const { url, html, waitForNetworkIdle = false } = args;
+
+  if (!url && !html) throw new Error("必须提供 url 或 html 参数");
+  const { targetUrl, originalUrl, rewritten } = await resolveCfBrowserUrl(url);
+
+  const gotoOptions = waitForNetworkIdle ? { waitUntil: "networkidle0" } : undefined;
+
+  const data = await callToolApi<{
+    markdown: string;
+    success: boolean;
+    browserMsUsed?: number;
+    source: string;
+  }>(thunkApi, "/api/cf-markdown", { url: targetUrl, html, gotoOptions }, { withAuth: true });
+
+  const extractionIssue = detectExtractionIssue(data.markdown ?? "", targetUrl ?? url ?? "网页");
+  if (extractionIssue) {
+    throw new Error(extractionIssue.message);
+  }
+
+  const seconds = data.browserMsUsed ? (data.browserMsUsed / 1000).toFixed(2) : "?";
+
+  return {
+    rawData: rewritten ? { ...data, originalUrl, resolvedUrl: targetUrl } : data,
+    displayData:
+      `📄 Markdown 提取完成\n- 来源: ${data.source}\n- 浏览器耗时: ${seconds}s\n- 字符数: ${data.markdown?.length ?? 0}` +
+      buildNormalizationNote(originalUrl, targetUrl) +
+      `\n\n` +
+      data.markdown,
+  };
+}
+
+// ──────────────────────────────────────────────────────────
+// cfGeneratePDF
+// ──────────────────────────────────────────────────────────
+export const cfGeneratePDFFunctionSchema = {
+  name: "cfGeneratePDF",
+  description:
+    "使用 Cloudflare Browser Rendering 将网页或 HTML 渲染为 PDF 文件。" +
+    "适合导出文档、报告、打印版网页等场景。",
+  parameters: {
+    type: "object",
+    properties: {
+      url: {
+        type: "string",
+        description: "要转换为 PDF 的网页 URL（http/https），与 html 二选一。",
+      },
+      html: {
+        type: "string",
+        description: "直接传入 HTML 字符串转换为 PDF，与 url 二选一。",
+      },
+      pdfOptions: {
+        type: "object",
+        description: "PDF 生成选项，例如 { format: 'A4', landscape: false, printBackground: true }。",
+        properties: {
+          format: { type: "string", description: "页面大小，如 'A4'、'Letter'。" },
+          landscape: { type: "boolean" },
+          printBackground: { type: "boolean" },
+        },
+      },
+    },
+  },
+};
+
+export async function cfGeneratePDFFunc(
+  args: {
+    url?: string;
+    html?: string;
+    pdfOptions?: { format?: string; landscape?: boolean; printBackground?: boolean };
+  },
+  thunkApi: any
+): Promise<{ rawData: any; displayData: string }> {
+  const { url, html, pdfOptions } = args;
+
+  if (!url && !html) throw new Error("必须提供 url 或 html 参数");
+  const { targetUrl, originalUrl, rewritten } = await resolveCfBrowserUrl(url);
+
+  const data = await callToolApi<{
+    dataUrl: string;
+    mimeType: string;
+    filename: string;
+    browserMsUsed?: number;
+    source: string;
+  }>(thunkApi, "/api/cf-pdf", { url: targetUrl, html, pdfOptions }, { withAuth: true });
+
+  const seconds = data.browserMsUsed ? (data.browserMsUsed / 1000).toFixed(2) : "?";
+
+  return {
+    rawData: rewritten ? { ...data, originalUrl, resolvedUrl: targetUrl } : data,
+    displayData:
+      `📑 PDF 生成完成\n- 来源: ${data.source}\n- 文件名: ${data.filename}\n- 浏览器耗时: ${seconds}s` +
+      buildNormalizationNote(originalUrl, targetUrl) +
+      `\n\n` +
+      `[下载 PDF](${data.dataUrl})`,
+  };
+}
+
+// ──────────────────────────────────────────────────────────
+// cfExtractJSON
+// ──────────────────────────────────────────────────────────
+export const cfExtractJSONFunctionSchema = {
+  name: "cfExtractJSON",
+  description:
+    "使用 Cloudflare Browser Rendering + AI，从网页中提取结构化数据。" +
+    "用自然语言描述你想要的数据（prompt），或提供 JSON Schema 定义结构，AI 自动抓取并返回结构化 JSON。" +
+    "适合：商品信息、职位列表、文章元数据、价格比较、任何需要从网页提取结构化数据的场景。",
+  parameters: {
+    type: "object",
+    properties: {
+      url: {
+        type: "string",
+        description: "目标网页 URL（http/https），与 html 二选一。",
+      },
+      html: {
+        type: "string",
+        description: "直接传入 HTML 字符串，与 url 二选一。",
+      },
+      prompt: {
+        type: "string",
+        description:
+          "用自然语言描述要提取的数据，例如：'提取所有商品的名称、价格和库存状态'。" +
+          "与 response_format 至少提供一个，同时提供时 prompt 引导提取方向，response_format 约束结构。",
+      },
+      response_format: {
+        type: "object",
+        description:
+          "JSON Schema 格式的结构定义，用于约束返回的数据结构。" +
+          "格式: { type: 'json_schema', schema: { type: 'object', properties: {...} } }",
+      },
+    },
+    required: [],
+  },
+};
+
+export async function cfExtractJSONFunc(
+  args: {
+    url?: string;
+    html?: string;
+    prompt?: string;
+    response_format?: object;
+  },
+  thunkApi: any
+): Promise<{ rawData: any; displayData: string }> {
+  const { url, html, prompt, response_format } = args;
+
+  if (!url && !html) throw new Error("必须提供 url 或 html 参数");
+  if (!prompt && !response_format) throw new Error("必须提供 prompt 或 response_format");
+  const { targetUrl, originalUrl, rewritten } = await resolveCfBrowserUrl(url);
+
+  const data = await callToolApi<{
+    result: unknown;
+    success: boolean;
+    browserMsUsed?: number;
+    source: string;
+  }>(
+    thunkApi,
+    "/api/cf-json",
+    { url: targetUrl, html, prompt, response_format },
+    { withAuth: true }
+  );
+
+  const seconds = data.browserMsUsed ? (data.browserMsUsed / 1000).toFixed(2) : "?";
+  const resultStr = JSON.stringify(data.result, null, 2);
+
+  return {
+    rawData: rewritten ? { ...data, originalUrl, resolvedUrl: targetUrl } : data,
+    displayData:
+      `🔍 结构化数据提取完成\n- 来源: ${data.source}\n- 浏览器耗时: ${seconds}s` +
+      buildNormalizationNote(originalUrl, targetUrl) +
+      `\n\n` +
+      `\`\`\`json\n${resultStr}\n\`\`\``,
+  };
+}

package/ai/tools/cfSpeechToTextTool.ts

@@ -0,0 +1,49 @@
+// packages/ai/tools/cfSpeechToTextTool.ts
+// Cloudflare Workers AI Whisper 语音转文字工具
+// Docs: https://developers.cloudflare.com/workers-ai/models/whisper/
+
+import { callToolApi } from "./toolApiClient";
+
+export const cfSpeechToTextFunctionSchema = {
+  name: "cfSpeechToText",
+  description:
+    "使用 Cloudflare Workers AI (@cf/openai/whisper) 将音频文件转换为文字。" +
+    "支持多种语言，自动识别语言。接受公开可访问的音频 URL。" +
+    "价格：$0.00045/分钟，适合偶发的语音转写任务。",
+  parameters: {
+    type: "object",
+    properties: {
+      audioUrl: {
+        type: "string",
+        description: "音频文件的公开访问 URL（支持 mp3、wav、flac、ogg 等格式）。",
+      },
+    },
+    required: ["audioUrl"],
+  },
+};
+
+export async function cfSpeechToTextFunc(
+  args: { audioUrl: string },
+  thunkApi: any
+): Promise<{ rawData: any; displayData: string }> {
+  const { audioUrl } = args;
+
+  if (!audioUrl || typeof audioUrl !== "string") {
+    throw new Error("必须提供有效的 audioUrl");
+  }
+
+  const data = await callToolApi<{
+    text: string;
+    wordCount?: number;
+    words?: Array<{ word: string; start: number; end: number }>;
+    vtt?: string;
+    source: string;
+  }>(thunkApi, "/api/cf-speech-to-text", { audioUrl }, { withAuth: true });
+
+  return {
+    rawData: data,
+    displayData:
+      `🎤 语音转文字完成\n- 来源: ${data.source}\n- 字数: ${data.wordCount ?? "?"}\n\n` +
+      `**转录内容：**\n\n${data.text}`,
+  };
+}

package/ai/tools/checkEnvTool.ts

@@ -0,0 +1,102 @@
+// 文件路径: packages/ai/tools/checkEnvTool.ts
+// 用于在本地环境执行“生效前检查”（当前最小实现：build）
+
+import { callToolApi } from "./toolApiClient";
+
+export type CheckEnvArgs = {
+    check?: "build" | "context";
+    key?: string;
+};
+
+export const checkEnvFunctionSchema = {
+    name: "checkEnv",
+    description: [
+        "执行环境检查并返回结构化结果。",
+        "check='context' 可查看当前平台、工作目录与可用 shell；check='build' 可快速验证是否可编译。",
+        "key 可检查某个本地工具，例如 key='gemini' 会探测本机 Gemini CLI 与 npm 最新版本。",
+    ].join("\n"),
+    parameters: {
+        type: "object",
+        properties: {
+            check: {
+                type: "string",
+                enum: ["build", "context"],
+                description: "检查项：context 返回当前运行环境；build 执行构建检查。",
+            },
+            key: {
+                type: "string",
+                enum: ["gemini", "node", "npm", "bun", "python", "git", "shell"],
+                description: "可选：探测某个本地工具，不执行 build。常用值如 gemini、node、npm、bun、python、git、shell。",
+            },
+        },
+    },
+};
+
+const formatProbeOutput = (probe: any): string => {
+    const status = probe?.ok ? "OK" : "FAIL";
+    const lines = [`- [${status}] ${probe?.label ?? "probe"}`];
+    const stdout = typeof probe?.stdout === "string" ? probe.stdout.trim() : "";
+    const stderr = typeof probe?.stderr === "string" ? probe.stderr.trim() : "";
+    const error = typeof probe?.error === "string" ? probe.error.trim() : "";
+    if (stdout) lines.push(`  stdout: ${stdout}`);
+    if (stderr) lines.push(`  stderr: ${stderr}`);
+    if (error) lines.push(`  error: ${error}`);
+    return lines.join("\n");
+};
+
+export async function checkEnvFunc(
+    args: CheckEnvArgs,
+    thunkApi: any,
+    context?: { agentKey?: string }
+): Promise<{ rawData: any; displayData: string }> {
+    const key = typeof args?.key === "string" ? args.key.trim().toLowerCase() : "";
+    const check = args?.check === "context" ? "context" : "build";
+    const payload = key ? { key } : { check };
+
+    const result = await callToolApi<any>(
+        thunkApi,
+        "/api/check-env",
+        payload,
+        {
+            withAuth: true,
+            agentKey: context?.agentKey,
+        }
+    );
+
+    const ok = !!result?.ok;
+    const exitCode = typeof result?.exitCode === "number" ? result.exitCode : -1;
+
+    if (result?.check === "tool") {
+        const probes = Array.isArray(result?.probes) ? result.probes : [];
+        return {
+            rawData: result,
+            displayData: [
+                `本地工具检查: ${result?.key ?? key}`,
+                ...probes.map(formatProbeOutput),
+            ].join("\n"),
+        };
+    }
+
+    if (check === "context") {
+        const displayParts = [
+            `平台: ${result?.platform ?? "unknown"}`,
+            `cwd: ${result?.cwd ?? "unknown"}`,
+            `defaultShell: ${result?.defaultShell ?? "unknown"}`,
+            `bash: ${result?.shells?.bash ? "yes" : "no"}`,
+            `powershell: ${result?.shells?.powershell ? "yes" : "no"}`,
+            `pwsh: ${result?.shells?.pwsh ? "yes" : "no"}`,
+        ];
+
+        return {
+            rawData: result,
+            displayData: `🧭 当前环境\n${displayParts.join("\n")}`,
+        };
+    }
+
+    return {
+        rawData: result,
+        displayData: ok
+            ? `✅ 环境检查通过: ${check}`
+            : `❌ 环境检查失败: ${check} (exitCode=${exitCode})`,
+    };
+}