openclaw-server 0.1.0 → 0.2.0

package/src/bookmark-search/parser.ts

@@ -0,0 +1,235 @@
+import type { BookmarkSearchMessageInspection } from "./types.js";
+
+const SEARCH_COMMAND_PATTERNS = [
+  /^(?:请|麻烦|帮我|给我|我想|我要|想|帮忙)?\s*(?:在)?(?:书签篮|书签栏|书签蓝)?(?:里)?\s*(?:搜索|搜一下|搜下|搜个|搜搜|搜)\s*[:：]?\s*(.+)$/iu,
+  /^(?:请|麻烦|帮我|给我|我想|我要|想|帮忙)?\s*(?:在)?(?:书签篮|书签栏|书签蓝)?(?:里)?\s*(?:查找|查询)\s*[:：]?\s*(.+)$/iu,
+  /^(?:请|麻烦|帮我|给我|我想|我要|想|帮忙)?\s*(?:在)?(?:书签篮|书签栏|书签蓝)?(?:里)?\s*(?:找一下|找找|找篇|找一篇|找)\s*[:：]?\s*(.+)$/iu,
+  /^(?:search|find|look up)\s+(.+)$/iu,
+];
+
+const LATEST_COMMAND_PATTERNS = [
+  /^(?:请|麻烦|帮我|给我|我想|我要|想|帮忙)?\s*(?:在)?(?:书签篮|书签栏|书签蓝)?(?:里)?\s*(?:看看|查看|列出|展示|打开|来点)?\s*(?:最新更新|最近更新|最新文章|最近文章|全部文章|最新资源|最近资源)\s*$/iu,
+];
+
+const CATEGORY_LIST_PATTERNS = [
+  /^(?:请|麻烦|帮我|给我|我想|我要|想|帮忙)?\s*(?:在)?(?:书签篮|书签栏|书签蓝)?(?:里)?\s*(?:看看|查看|列出|展示|打开)?\s*(?:分类|分类列表|分类目录|分类导航|一级分类|一级分类列表)\s*$/iu,
+  /^(?:在)?(?:书签篮|书签栏|书签蓝)?(?:里)?(?:有|有哪些)(?:什么)?(?:分类|一级分类)\s*$/iu,
+];
+
+const CATEGORY_LINK_PATTERNS = [
+  /^(?:请|麻烦|帮我|给我|我想|我要|想|帮忙)?\s*(?:在)?(?:书签篮|书签栏|书签蓝)?(?:里)?\s*(?:看看|查看|列出|展示|打开)?\s*(?:分类|栏目|目录)\s*[:：]?\s*(.+?)\s*(?:的)?(?:常用链接|常用网站|常用站点|常用网址)\s*$/iu,
+  /^(?:请|麻烦|帮我|给我|我想|我要|想|帮忙)?\s*(?:在)?(?:书签篮|书签栏|书签蓝)?(?:里)?\s*(?:看看|查看|列出|展示|打开)?\s*(.+?)\s*(?:分类|栏目|目录)?(?:的)?(?:常用链接|常用网站|常用站点|常用网址)\s*$/iu,
+];
+
+const CATEGORY_ARTICLE_PATTERNS = [
+  /^(?:请|麻烦|帮我|给我|我想|我要|想|帮忙)?\s*(?:在)?(?:书签篮|书签栏|书签蓝)?(?:里)?\s*(?:看看|查看|列出|展示|打开|推荐)?\s*(?:分类|栏目|目录)\s*[:：]?\s*(.+?)\s*(?:的)?(?:文章|资源|内容)?\s*$/iu,
+  /^(?:请|麻烦|帮我|给我|我想|我要|想|帮忙)?\s*(?:在)?(?:书签篮|书签栏|书签蓝)?(?:里)?\s*(?:看看|查看|列出|展示|打开|推荐)?\s*(.+?)\s*(?:分类|栏目|目录)\s*(?:的)?(?:文章|资源|内容)?\s*$/iu,
+];
+
+const SEARCH_HINT_PATTERN =
+  /(?:书签篮|书签栏|书签蓝|搜索|搜一下|搜下|搜个|搜搜|查找|查询|search|find|look up)/iu;
+
+const IMPLICIT_SEARCH_EXACT_BLOCKLIST = new Set([
+  "你好",
+  "您好",
+  "嗨",
+  "哈喽",
+  "hello",
+  "hi",
+  "hey",
+  "在吗",
+  "帮助",
+  "help",
+  "谢谢",
+  "thanks",
+  "thank you",
+  "再见",
+  "bye",
+]);
+
+const IMPLICIT_SEARCH_PREFIX_BLOCK_PATTERN =
+  /^(?:你|我|请|帮我|麻烦|怎么|如何|为什么|什么|谁|哪|几|能不能|可以|会不会|有没有|要不要|是否|介绍一下|说一下|告诉我)/iu;
+
+const IMPLICIT_SEARCH_QUESTION_PATTERN = /(?:如何|怎么样|怎样|咋样|吗|呢|么|嘛|\?|？)/iu;
+
+const EMPTY_QUERY_TOKENS = new Set([
+  "书签",
+  "书签篮",
+  "书签栏",
+  "书签蓝",
+  "文章",
+  "分类",
+  "结果",
+  "内容",
+  "链接",
+  "网址",
+  "资料",
+  "一下",
+]);
+
+function cleanupFreeText(query: string): string {
+  return query
+    .normalize("NFKC")
+    .replace(/^[\s"'`“”‘’《》【】（）()]+|[\s"'`“”‘’《》【】（）()]+$/gu, "")
+    .replace(/^(?:关于|有关|相关|相关文章|相关结果|相关内容|分类|栏目|目录)\s*/iu, "")
+    .replace(/\s*(?:相关的)?(?:书签|文章|结果|内容|链接|网址|资料)\s*$/iu, "")
+    .replace(/\s+/g, " ")
+    .trim();
+}
+
+function cleanupSearchQuery(query: string): string {
+  return cleanupFreeText(query);
+}
+
+function cleanupCategoryName(query: string): string {
+  return cleanupFreeText(query)
+    .replace(/\s*(?:分类|栏目|目录)\s*$/iu, "")
+    .trim();
+}
+
+function isMeaningfulSearchQuery(query: string): boolean {
+  const normalized = cleanupSearchQuery(query);
+  if (!normalized) {
+    return false;
+  }
+
+  return !EMPTY_QUERY_TOKENS.has(normalized.toLowerCase());
+}
+
+function extractImplicitBookmarkSearchQuery(text: string): string | undefined {
+  const query = cleanupSearchQuery(text);
+  if (!isMeaningfulSearchQuery(query)) {
+    return undefined;
+  }
+
+  const normalized = query.toLowerCase();
+  if (IMPLICIT_SEARCH_EXACT_BLOCKLIST.has(normalized)) {
+    return undefined;
+  }
+
+  if (IMPLICIT_SEARCH_PREFIX_BLOCK_PATTERN.test(query)) {
+    return undefined;
+  }
+
+  if (IMPLICIT_SEARCH_QUESTION_PATTERN.test(query)) {
+    return undefined;
+  }
+
+  if (query.includes("\n") || query.length > 40) {
+    return undefined;
+  }
+
+  const wordCount = query.split(/\s+/u).filter(Boolean).length;
+  if (wordCount > 6) {
+    return undefined;
+  }
+
+  return query;
+}
+
+export function extractBookmarkSearchQuery(text: string): string | undefined {
+  for (const pattern of SEARCH_COMMAND_PATTERNS) {
+    const match = text.match(pattern);
+    if (!match?.[1]) {
+      continue;
+    }
+
+    const query = cleanupSearchQuery(match[1]);
+    if (isMeaningfulSearchQuery(query)) {
+      return query;
+    }
+    return undefined;
+  }
+
+  return undefined;
+}
+
+function extractBookmarkCategoryName(
+  patterns: RegExp[],
+  text: string,
+): string | undefined {
+  for (const pattern of patterns) {
+    const match = text.match(pattern);
+    if (!match?.[1]) {
+      continue;
+    }
+
+    const category = cleanupCategoryName(match[1]);
+    if (category) {
+      return category;
+    }
+  }
+
+  return undefined;
+}
+
+export function inspectBookmarkSearchQuery(text: string): BookmarkSearchMessageInspection {
+  const explicitSearchQuery = extractBookmarkSearchQuery(text);
+  if (explicitSearchQuery) {
+    return {
+      shouldHandle: true,
+      reason: "query",
+      action: "search",
+      query: explicitSearchQuery,
+    };
+  }
+
+  if (LATEST_COMMAND_PATTERNS.some((pattern) => pattern.test(text))) {
+    return {
+      shouldHandle: true,
+      reason: "query",
+      action: "latest",
+    };
+  }
+
+  if (CATEGORY_LIST_PATTERNS.some((pattern) => pattern.test(text))) {
+    return {
+      shouldHandle: true,
+      reason: "query",
+      action: "categories",
+    };
+  }
+
+  const categoryForLinks = extractBookmarkCategoryName(CATEGORY_LINK_PATTERNS, text);
+  if (categoryForLinks) {
+    return {
+      shouldHandle: true,
+      reason: "query",
+      action: "category_links",
+      category: categoryForLinks,
+    };
+  }
+
+  const categoryForArticles = extractBookmarkCategoryName(CATEGORY_ARTICLE_PATTERNS, text);
+  if (categoryForArticles) {
+    return {
+      shouldHandle: true,
+      reason: "query",
+      action: "category_articles",
+      category: categoryForArticles,
+    };
+  }
+
+  const query = extractImplicitBookmarkSearchQuery(text);
+  if (query) {
+    return {
+      shouldHandle: true,
+      reason: "query",
+      action: "search",
+      query,
+    };
+  }
+
+  if (SEARCH_HINT_PATTERN.test(text)) {
+    return {
+      shouldHandle: true,
+      reason: "query",
+      missing: "query",
+    };
+  }
+
+  return {
+    shouldHandle: false,
+    reason: "no_match",
+  };
+}

package/src/bookmark-search/service.test.ts

@@ -0,0 +1,330 @@
+import { describe, expect, it, vi } from "vitest";
+import { BookmarkSearchService } from "./service.js";
+import type { BookmarkCategory, BookmarkLink, BookmarkSearchProvider } from "./types.js";
+
+type ProviderSearchRecord = {
+  title: string;
+  type?: string;
+  desc?: string;
+  url: string;
+  articleCount?: number;
+  firstCategory?: string;
+  secondCategory?: string;
+  updatedAt?: string;
+};
+
+function createProvider(
+  records: Record<string, ProviderSearchRecord[]>,
+  extras: {
+    latestArticles?: ProviderSearchRecord[];
+    latestBrowseUrl?: string;
+    categories?: BookmarkCategory[];
+    categoryArticles?: Record<string, ProviderSearchRecord[]>;
+    categoryLinks?: Record<string, BookmarkLink[]>;
+  } = {},
+): BookmarkSearchProvider {
+  const toItem = (item: ProviderSearchRecord) => ({
+    articleCount: item.articleCount,
+    desc: item.desc,
+    firstCategory: item.firstCategory,
+    secondCategory: item.secondCategory,
+    type: item.type ?? "文章",
+    title: item.title,
+    updatedAt: item.updatedAt,
+    url: item.url,
+  });
+
+  return {
+    searchKeyword: vi.fn(async ({ keyword }) =>
+      (records[keyword] ?? []).map((item) => toItem(item)),
+    ),
+    listLatestArticles: vi.fn(async () => ({
+      items: (extras.latestArticles ?? []).map((item) => toItem(item)),
+      total: extras.latestArticles?.length ?? 0,
+      browseUrl: extras.latestBrowseUrl,
+    })),
+    listTopCategories: vi.fn(async () => extras.categories ?? []),
+    listCategoryArticles: vi.fn(async ({ category }) => ({
+      items: (extras.categoryArticles?.[category.name] ?? []).map((item) => toItem(item)),
+      total: extras.categoryArticles?.[category.name]?.length ?? 0,
+      browseUrl: category.url,
+    })),
+    listCategoryLinks: vi.fn(async ({ category }) => extras.categoryLinks?.[category.name] ?? []),
+  };
+}
+
+describe("BookmarkSearchService", () => {
+  it("renders only article results with linked titles and plain descriptions", async () => {
+    const service = new BookmarkSearchService(
+      createProvider({
+        python: [
+          {
+            title: "Python 教程：循环语句",
+            desc: "整理了 Python 循环语句、break 和 continue 的常见用法。",
+            type: "文章",
+            url: "https://shuqianlan.com/static-article/detail-html/python-loop.html",
+          },
+          {
+            title: "Python 资源合集",
+            type: "二级分类",
+            articleCount: 8,
+            url: "https://shuqianlan.com/static-article/cate-page/second/python.html",
+          },
+        ],
+      }),
+    );
+
+    const reply = await service.processMessage({ userId: "u1", text: "搜 python" });
+    expect(reply.intent).toBe("search_results");
+    expect(reply.reply).toContain("找到 1 条“python”相关结果");
+    expect(reply.reply).toContain("[Python 教程：循环语句](https://shuqianlan.com/static-article/detail-html/python-loop.html)");
+    expect(reply.reply).toContain("简介：整理了 Python 循环语句、break 和 continue 的常见用法。");
+    expect(reply.reply).not.toContain("[整理了 Python 循环语句、break 和 continue 的常见用法。]");
+    expect(reply.reply).not.toContain("文章");
+    expect(reply.reply).not.toContain("Python 资源合集");
+    expect(reply.reply).not.toContain("cate-page/second/python.html");
+  });
+
+  it("supports bare-keyword bookmark searches", async () => {
+    const service = new BookmarkSearchService(
+      createProvider({
+        百度: [
+          {
+            title: "百度一下，你就知道",
+            desc: "全球领先的中文搜索引擎。",
+            url: "https://shuqianlan.com/static-article/detail-html/baidu.html",
+          },
+        ],
+      }),
+    );
+
+    const reply = await service.processMessage({ userId: "u1", text: "百度" });
+    expect(reply.intent).toBe("search_results");
+    expect(reply.reply).toContain("百度一下，你就知道");
+    expect(reply.reply).toContain("全球领先的中文搜索引擎。");
+  });
+
+  it("falls back to token search and keeps the strongest intersection matches", async () => {
+    const service = new BookmarkSearchService(
+      createProvider({
+        notion: [
+          {
+            title: "Notion",
+            url: "https://shuqianlan.com/static-article/detail-html/notion.html",
+          },
+          {
+            title: "Notion AI",
+            url: "https://shuqianlan.com/static-article/detail-html/notion-ai.html",
+          },
+        ],
+        ai: [
+          {
+            title: "Notion AI",
+            url: "https://shuqianlan.com/static-article/detail-html/notion-ai.html",
+          },
+          {
+            title: "Midjourney",
+            url: "https://shuqianlan.com/static-article/detail-html/midjourney.html",
+          },
+        ],
+      }),
+    );
+
+    const reply = await service.processMessage({ userId: "u2", text: "搜索 notion ai" });
+    expect(reply.intent).toBe("search_results");
+    expect(reply.reply).toContain("Notion AI");
+    expect(reply.reply).not.toContain("Midjourney");
+  });
+
+  it("falls back to token search when exact matches only contain categories", async () => {
+    const service = new BookmarkSearchService(
+      createProvider({
+        "cesium tutorial": [
+          {
+            title: "Cesium 教程分类",
+            type: "二级分类",
+            articleCount: 5,
+            url: "https://shuqianlan.com/static-article/cate-page/second/cesium.html",
+          },
+        ],
+        cesium: [
+          {
+            title: "Cesium 获取地形高度 全面深度教程",
+            desc: "覆盖地形高度采样、坐标转换和实际项目中的精度处理。",
+            type: "文章",
+            url: "https://shuqianlan.com/static-article/detail-html/cesium-height.html",
+          },
+        ],
+        tutorial: [
+          {
+            title: "Cesium 获取地形高度 全面深度教程",
+            desc: "覆盖地形高度采样、坐标转换和实际项目中的精度处理。",
+            type: "文章",
+            url: "https://shuqianlan.com/static-article/detail-html/cesium-height.html",
+          },
+        ],
+      }),
+    );
+
+    const reply = await service.processMessage({ userId: "u2", text: "搜索 cesium tutorial" });
+    expect(reply.intent).toBe("search_results");
+    expect(reply.reply).toContain("Cesium 获取地形高度 全面深度教程");
+    expect(reply.reply).not.toContain("Cesium 教程分类");
+  });
+
+  it("numbers multiple article results while keeping only titles clickable", async () => {
+    const service = new BookmarkSearchService(
+      createProvider({
+        react: [
+          {
+            title: "React 状态管理清单",
+            desc: "梳理 useState、context 和状态拆分时的常见取舍与边界。",
+            type: "文章",
+            url: "https://shuqianlan.com/static-article/detail-html/react-state.html",
+          },
+          {
+            title: "React 性能排查指南",
+            desc: "整理定位重渲染、异步更新和列表卡顿时最常用的排查步骤。",
+            type: "文章",
+            url: "https://shuqianlan.com/static-article/detail-html/react-performance.html",
+          },
+        ],
+      }),
+    );
+
+    const reply = await service.processMessage({ userId: "u7", text: "搜 react" });
+    expect(reply.intent).toBe("search_results");
+    expect(reply.reply).toContain("1、[React 状态管理清单](https://shuqianlan.com/static-article/detail-html/react-state.html)");
+    expect(reply.reply).toContain("2、[React 性能排查指南](https://shuqianlan.com/static-article/detail-html/react-performance.html)");
+    expect(reply.reply).toContain(
+      "1、[React 状态管理清单](https://shuqianlan.com/static-article/detail-html/react-state.html)  \n简介：梳理 useState、context 和状态拆分时的常见取舍与边...",
+    );
+    expect(reply.reply).toContain(
+      "简介：梳理 useState、context 和状态拆分时的常见取舍与边...\n\n2、[React 性能排查指南](https://shuqianlan.com/static-article/detail-html/react-performance.html)",
+    );
+    expect(reply.reply).toContain("简介：梳理 useState、context 和状态拆分时的常见取舍与边...");
+    expect(reply.reply).not.toContain("[梳理 useState、context 和状态拆分时的常见取舍与边...");
+  });
+
+  it("lists latest updated articles from the bookmark directory", async () => {
+    const service = new BookmarkSearchService(
+      createProvider(
+        {},
+        {
+          latestArticles: [
+            {
+              title: "Cesium 获取地形高度 全面深度教程",
+              desc: "覆盖地形高度采样、坐标转换和实际项目中的精度处理。",
+              firstCategory: "开发工具",
+              secondCategory: "地图",
+              updatedAt: "2026-03-13 18:20:00",
+              url: "https://shuqianlan.com/static-article/detail-html/cesium-height.html",
+            },
+          ],
+          latestBrowseUrl: "https://shuqianlan.com/static-article/page/page_1.html",
+        },
+      ),
+    );
+
+    const reply = await service.processMessage({ userId: "u3", text: "最新文章" });
+    expect(reply.intent).toBe("latest_results");
+    expect(reply.reply).toContain("最近更新");
+    expect(reply.reply).toContain(
+      "[Cesium 获取地形高度 全面深度教程](https://shuqianlan.com/static-article/detail-html/cesium-height.html)",
+    );
+    expect(reply.reply).not.toContain("开发工具 > 地图");
+    expect(reply.reply).not.toContain("更新：2026-03-13 18:20:00");
+    expect(reply.reply).toContain("[查看全部文章](https://shuqianlan.com/static-article/page/page_1.html)");
+  });
+
+  it("lists first-level categories from the bookmark directory", async () => {
+    const service = new BookmarkSearchService(
+      createProvider(
+        {},
+        {
+          categories: [
+            {
+              name: "开发工具",
+              articleCount: 12,
+              url: "https://shuqianlan.com/static-article/cate-page/first/dev-tools.html",
+            },
+            {
+              name: "设计灵感",
+              articleCount: 8,
+              url: "https://shuqianlan.com/static-article/cate-page/first/design.html",
+            },
+          ],
+        },
+      ),
+    );
+
+    const reply = await service.processMessage({ userId: "u4", text: "分类列表" });
+    expect(reply.intent).toBe("category_list");
+    expect(reply.reply).toContain("一级分类");
+    expect(reply.reply).toContain("1、[开发工具](https://shuqianlan.com/static-article/cate-page/first/dev-tools.html)（12 篇）");
+    expect(reply.reply).toContain("前端分类");
+  });
+
+  it("shows category articles and common links", async () => {
+    const service = new BookmarkSearchService(
+      createProvider(
+        {},
+        {
+          categories: [
+            {
+              name: "开发工具",
+              articleCount: 12,
+              url: "https://shuqianlan.com/static-article/cate-page/first/dev-tools.html",
+            },
+          ],
+          categoryArticles: {
+            开发工具: [
+              {
+                title: "Vite 构建优化指南",
+                desc: "整理了依赖预构建、拆包和生产构建性能优化的常见做法。",
+                firstCategory: "开发工具",
+                secondCategory: "前端工程化",
+                updatedAt: "2026-03-12 09:30:00",
+                url: "https://shuqianlan.com/static-article/detail-html/vite-guide.html",
+              },
+            ],
+          },
+          categoryLinks: {
+            开发工具: [
+              {
+                name: "Vite",
+                url: "https://vite.dev",
+              },
+              {
+                name: "TypeScript",
+                url: "https://www.typescriptlang.org",
+              },
+            ],
+          },
+        },
+      ),
+    );
+
+    const categoryReply = await service.processMessage({ userId: "u5", text: "开发工具分类" });
+    expect(categoryReply.intent).toBe("category_results");
+    expect(categoryReply.reply).toContain("“开发工具”文章");
+    expect(categoryReply.reply).toContain("开发工具 > 前端工程化 · 更新：2026-03-12 09:30:00");
+    expect(categoryReply.reply).toContain("[打开“开发工具”分类页](https://shuqianlan.com/static-article/cate-page/first/dev-tools.html)");
+
+    const linksReply = await service.processMessage({ userId: "u6", text: "开发工具常用链接" });
+    expect(linksReply.intent).toBe("category_links");
+    expect(linksReply.reply).toContain("1、[Vite](https://vite.dev)");
+    expect(linksReply.reply).toContain("2、[TypeScript](https://www.typescriptlang.org)");
+  });
+
+  it("asks for a query when the user only says to search", async () => {
+    const service = new BookmarkSearchService(createProvider({}));
+
+    const firstReply = await service.processMessage({ userId: "u3", text: "帮我搜一下" });
+    expect(firstReply.intent).toBe("clarify");
+
+    const secondReply = await service.processMessage({ userId: "u3", text: "cesium" });
+    expect(secondReply.intent).toBe("not_found");
+    expect(secondReply.reply).toContain("cesium");
+  });
+});