Files
Interview-Notebook/website/builder/test_search.cjs
T
2026-11-07 05:34:06 +08:00

149 lines
12 KiB
JavaScript
Raw Blame History

This file contains ambiguous Unicode characters
This file contains Unicode characters that might be confused with other characters. If you think that this is intentional, you can safely ignore this warning. Use the Escape button to reveal them.
const assert = require('node:assert/strict');
const fs = require('node:fs');
const path = require('node:path');
const { search, highlight } = require('../templates/search.js');
const dist = process.argv[2] || path.join(__dirname, '../dist');
const info = JSON.parse(fs.readFileSync(path.join(dist, 'build-info.json')));
const index = JSON.parse(fs.readFileSync(path.join(dist, info.searchIndex)));
assert(search(index, '拥塞控制').some(r => r.path.includes('计算机网络 - 传输层') && r.id === 'tcp-拥塞控制'));
assert(search(index, '检索增强生成').some(r => r.path === '/notes/大模型应用' && r.id === '三、检索增强生成'));
assert(search(index, 'Token 分词').some(r => r.path === '/notes/大模型应用' && r.id === 'token-与分词'));
assert.equal(search(index, '').length, 0);
const ranking = { documents: [
{ title: '学习路线 - 示例', path: '/notes/学习路线 - 示例', sections: [{ title: '第 1 天:拥塞控制', id: 'day', text: '阅读拥塞控制' }] },
{ title: '网络', path: '/notes/网络', sections: [
{ title: 'TCP 拥塞控制', id: 'tcp', text: '拥塞控制用于……' },
{ title: 'A1 线程池的拒绝策略', id: 'a1', text: '线程池实验' },
{ title: 'ThreadPoolExecutor', id: 'pool', text: '线程池指的是……' },
{ title: 'Thread Local Storage', id: 'tls', text: '线程本地存储' },
{ title: 'Agentic RAG', id: 'rag', text: '检索' },
] },
] };
assert.equal(search(ranking, '拥塞控制')[0].id, 'tcp', '讲解正文排在学习路线之前');
assert.equal(search(ranking, '线程池')[0].id, 'pool', '讲解正文排在附录实验之前');
assert.equal(search(ranking, 'rag')[0].id, 'rag', '英文词按完整单词优先');
assert.equal(search(ranking, '第 1 天')[0].id, 'day', '学习路线仍能搜到');
const phrasing = { documents: [
{ title: '死锁', path: '/notes/死锁', sections: [
{ title: '死锁预防', id: 'pre', text: '破坏某个必要条件' },
{ title: '必要条件', id: 'cond', text: '只有以下四个条件同时成立时才会发生死锁' },
] },
{ title: '容器', path: '/notes/容器', sections: [
{ title: 'HashMap', id: 'map', text: '哈希冲突用拉链法解决,扩容时容量翻倍' },
{ title: '与 Hashtable 的比较', id: 'cmp', text: 'HashMap 不是线程安全的' },
] },
{ title: 'Java 虚拟机', path: '/notes/jvm', sections: [{ title: '垃圾收集器', id: 'gc', text: 'G1' }] },
{ title: '57.1 和为 S 的两个数字', path: '/notes/57.1', sections: [
{ title: '57.1 和为 S 的两个数字', id: '', text: '' }, { title: '57.1 和为 S 的两个数字', id: 'h1', text: '' },
] },
] };
assert.equal(search(phrasing, '死锁的必要条件')[0].id, 'cond', '去掉问法后,小节标题正好匹配的排在前面');
assert.equal(search(phrasing, '死锁的四个条件')[0].id, 'cond', '去掉问法后再搜');
assert(search(phrasing, '同时成立的四条件').fuzzy && search(phrasing, '同时成立的四条件')[0].id === 'cond', '没有完全匹配时按词组找相近的小节,并标出来');
assert.equal(search(phrasing, 'hashmap扩容')[0].id, 'map', '中英文连写时拆开');
assert.equal(search(phrasing, '散列冲突')[0].id, 'map', '术语的另一种说法');
assert.equal(search(phrasing, '垃圾回收器')[0].id, 'gc', '垃圾回收按垃圾收集搜索');
assert.equal(search(phrasing, 'HashMap和Hashtable的区别')[0].id, 'cmp', '“和……的区别”问法');
assert.equal(search(phrasing, '和为 S').length, 1, '没有正文的开头部分不和一级标题重复出现');
assert.equal(search(phrasing, '什么是').length, 0, '只有问法时没有结果');
assert.equal(highlight('a<b 的 TCP 与 tcp', 'tcp'), 'a&lt;b 的 <mark>TCP</mark> 与 <mark>tcp</mark>', '搜索词高亮不区分大小写,其余文字转义');
assert.equal(highlight('&amp; (x)', 'amp (x)'), '&amp;<mark>amp</mark>; <mark>(x)</mark>', '高亮先切分再转义,正则字符按字面匹配');
const generic = { documents: [
{ title: '学习路线 - 示例', path: '/notes/学习路线 - 示例', sections: [{ title: '第 3 天:锁', id: 'day', text: '阅读 synchronized 的原理' }] },
{ title: 'Spring', path: '/notes/Spring', sections: [
{ title: '七、Spring 中的设计模式', id: 'pattern', text: 'AOP 用到了代理模式' },
{ title: '二、AOP', id: 'aop', text: 'Spring AOP 在运行时为 Bean 生成代理' },
] },
{ title: '并发', path: '/notes/并发', sections: [{ title: 'synchronized', id: 'sync', text: 'JVM 用监视器锁实现' }] },
{ title: '大模型', path: '/notes/大模型', sections: [
{ title: '长上下文与 Agentic RAG', id: 'agentic', text: 'RAG 的变体' },
{ title: '三、检索增强生成', id: 'rag', text: '先检索再生成' },
] },
{ title: '进程管理', path: '/notes/进程管理', sections: [
{ title: '进程与线程', id: 'both', text: '进程是资源分配的单位,线程是调度的单位' },
{ title: '2. 线程', id: 'thread', text: '线程共享进程的资源' },
] },
] };
assert.equal(search(generic, 'synchronized 原理')[0].id, 'sync', '“原理”这类泛泛的词正文中没有时,去掉后再搜');
assert.equal(search(generic, 'spring aop')[0].id, 'aop', '文章名中的词出现在小节标题里少加分');
assert.equal(search(generic, 'rag 是什么')[0].id, 'rag', 'RAG 与检索增强生成互通,去掉编号后正好是搜索词的小节在前');
assert.equal(search(generic, '进程和线程的区别')[0].id, 'both', '多个词时,只匹配其中一个词的小节不算正好匹配');
const golden = { documents: [
{ title: '设计模式 - 单例', path: '/notes/单例', sections: [{ title: '单例(Singleton)', id: 'single', text: '确保一个类只有一个实例' }] },
{ title: '攻击技术', path: '/notes/攻击', sections: [{ title: '一、跨站脚本攻击', id: 'xss', text: '攻击者把脚本注入页面' }] },
{ title: '数据库', path: '/notes/库', sections: [
{ title: '四、隔离级别', id: 'level', text: '未提交读、提交读、可重复读、可串行化' },
{ title: '参考资料', id: 'refs', text: '事务隔离级别 的文档' },
] },
{ title: '容器 - 进阶', path: '/notes/容器 - 进阶', sections: [{ title: 'ConcurrentHashMap', id: 'deep', text: 'ConcurrentHashMap 的源码' }] },
{ title: '容器', path: '/notes/容器', sections: [{ title: 'ConcurrentHashMap', id: 'main', text: 'ConcurrentHashMap 的源码' }] },
] };
assert.equal(search(golden, '单例模式')[0].id, 'single', '“单例模式”也按“单例”搜');
assert.equal(search(golden, 'XSS')[0].id, 'xss', 'XSS 与跨站脚本互通');
assert.equal(search(golden, '事务隔离级别')[0].id, 'level', '完全匹配的只有参考资料时,相近的小节排在前面');
assert(!search(golden, '事务隔离级别').fuzzy, '有完全匹配的结果时不提示“没有完全匹配”');
assert.equal(search(golden, 'concurrenthashmap')[0].id, 'main', '主笔记排在进阶篇之前');
const natural = { documents: [
{ title: 'MySQL', path: '/notes/MySQL', sections: [
{ title: '索引失效', id: 'invalid', text: '对索引列做运算、函数或隐式类型转换时用不上索引' },
{ title: '索引的优点', id: 'good', text: '减少扫描的行数' },
{ title: '主从复制', id: 'repl', text: 'binlog 线程把日志发给从库' },
] },
{ title: '操作系统', path: '/notes/操作系统', sections: [{ title: '5. 协程', id: 'co', text: '用户态的轻量线程' }] },
] };
assert.equal(search(natural, '什么情况索引会失效?')[0].id, 'invalid', '整句的问法按词组的相关度找到小节');
assert.equal(search(natural, '平时是怎么用协程的')[0].id, 'co', '问法中的“平时”“用”这类词不影响');
assert.equal(search(natural, '主从复制')[0].id, 'repl', '短的搜索词仍按完整匹配排序');
const asking = { documents: [
{ title: 'Java 虚拟机', path: '/notes/Java 虚拟机', sections: [
{ title: '堆', id: 'heap', text: '所有对象都在这里分配内存,是垃圾收集的主要区域' },
{ title: '虚拟机栈', id: 'stack', text: '每个 Java 方法在执行的同时会创建一个栈帧' },
{ title: '堆和栈的区别', id: 'diff', text: '堆存放对象,线程共享;栈存放局部变量,线程私有' },
] },
{ title: 'Java 并发', path: '/notes/Java 并发', sections: [{ title: '七、J.U.C - AQS', id: 'aqs', text: 'java.util.concurrent 大大提高了并发性能' }] },
{ title: '算法', path: '/notes/算法', sections: [{ title: '跳表', id: 'skip', text: '多层有序链表' }] },
] };
assert.equal(search(asking, '堆和栈的区别是什么?')[0].id, 'diff', '去掉问句的开头结尾后正好是小节标题');
assert.equal(search(asking, '了解juc吗?')[0].id, 'aqs', 'JUC 与 J.U.C 互通');
assert.equal(search(asking, '跳跃表')[0].id, 'skip', '跳跃表与跳表互通');
const skiplist = { documents: [
{ title: 'Redis', path: '/notes/Redis', sections: [{ title: '跳跃表', id: 'skip', text: '有序集合的底层实现之一,插入时随机决定节点的层数' }] },
{ title: '攻击技术', path: '/notes/攻击技术', sections: [
{ title: '设置 Cookie 为 HttpOnly', id: 'cookie', text: '设置后脚本无法读取 Cookie' },
{ title: '设置 Cookie 的 SameSite 属性', id: 'samesite', text: '设置为 Strict 时跨站请求不携带 Cookie' },
] },
{ title: 'Linux', path: '/notes/Linux', sections: [{ title: '环境变量', id: 'env', text: '在配置文件中设置' }] },
{ title: 'MySQL', path: '/notes/MySQL', sections: [{ title: '参数', id: 'param', text: '设置缓冲池大小' }] },
] };
assert.equal(search(skiplist, '跳表的层高怎么设置')[0].id, 'skip', '整句里的“跳表”也能匹配标题“跳跃表”');
const camel = { documents: [
{ title: '微服务', path: '/notes/微服务', sections: [{ title: '二、Spring Cloud', id: 'sc', text: 'Spring Cloud Alibaba 提供 Nacos、Sentinel 等组件' }] },
{ title: '缓存', path: '/notes/缓存', sections: [{ title: '七、LRU', id: 'lru', text: '下面是基于 LinkedHashMap 实现的 LRU Cache' }] },
] };
assert.equal(search(camel, 'SpringCloudAlibaba')[0].id, 'sc', '驼峰连写按拆开的写法也能搜到');
assert.equal(search(camel, 'LRUCache')[0].id, 'lru', '连续大写后接单词时在单词前拆开');
const versions = { documents: [{ title: 'HTTP', path: '/notes/HTTP', sections: [
{ title: '八、HTTP/2.0', id: 'h2', text: '二进制分帧、多路复用、首部压缩' },
{ title: '七、HTTP/1.1 新特性', id: 'h11', text: '默认长连接' },
{ title: '一、基础概念', id: 'base', text: '请求和响应' },
] }] };
assert.equal(search(versions, 'HTTP2.0 有哪些优势?')[0].id, 'h2', '“HTTP2.0”按笔记中的写法“HTTP/2.0”搜');
assert.equal(search(versions, 'http2')[0].id, 'h2');
assert(search(versions, 'HTTP1.1、2.0 的区别').slice(0, 2).some(r => r.id === 'h11'), '顿号分开的版本号是两个词');
const sample = { documents: [{ title: '数据库', path: '/notes/示例', sections: [
{ title: '事务', id: '事务', text: '一致性 '.repeat(20) },
{ title: '一致性', id: '一致性', text: 'SQL 与数据库' },
] }] };
const serialized = JSON.stringify(sample);
assert.equal(search(sample, '一致性')[0].id, '一致性', '章节标题应优先于正文中重复的关键词');
assert.equal(search(sample, 'SQL')[0].id, '一致性', '全角输入仍能搜索');
assert.equal(JSON.stringify(sample), serialized, '搜索缓存不能改动原索引');
const nextIndex = JSON.parse(serialized);
nextIndex.documents[0].sections.push({ title: '新增内容', id: '新章节', text: '新笔记检索词' });
assert.equal(search(nextIndex, '新笔记检索词')[0].id, '新章节', '新索引应使用新内容');
assert.equal(search(sample, '新笔记检索词').length, 0, '不同索引的缓存不能串用');
if (process.argv.includes('--updated')) {
assert(search(index, '本地同步验证示例词').some(r => r.id === '本地验证更新'));
}
console.log(`搜索与锚点验证通过:${index.documents.length} 篇文章。`);