94 lines
4.5 KiB
JavaScript
94 lines
4.5 KiB
JavaScript
import { test } from 'node:test';
|
||
import assert from 'node:assert/strict';
|
||
import { readFileSync, readdirSync, existsSync } from 'node:fs';
|
||
|
||
// 计数散落多处，历史上漏同步过 5 处（见 CLAUDE.md「同步计数与导航」）。
|
||
// 本测试把那份人工清单变成检查：新技能增删后，任何一处没跟上都会红。
|
||
const SKILLS_DIR = '.claude/skills';
|
||
|
||
function actualCounts() {
|
||
let atomic = 0, gateway = 0, entry = 0;
|
||
let total = 0;
|
||
for (const d of readdirSync(SKILLS_DIR)) {
|
||
if (!d.startsWith('re-') || !existsSync(`${SKILLS_DIR}/${d}/SKILL.md`)) continue;
|
||
total++;
|
||
const m = /^type:\s*(\w+)/m.exec(readFileSync(`${SKILLS_DIR}/${d}/SKILL.md`, 'utf8'));
|
||
const kind = m ? m[1] : 'atomic';
|
||
if (kind === 'gateway') gateway++;
|
||
else if (kind === 'entry') entry++;
|
||
else atomic++;
|
||
}
|
||
return { total, atomic, gateway, entry };
|
||
}
|
||
|
||
// file → [正则, 期望口径, 说明]；正则第 1 个捕获组是要比对的数字
|
||
const SITES = [
|
||
['README.md', /^(\d+) 个逆向工程技能/m, 'total', '首段技能总数'],
|
||
['README.md', /^## 技能导航（(\d+)）/m, 'total', '技能导航标题'],
|
||
['README.md', /^入口 → (\d+) 大类网关 → (\d+) 原子技能/m, 'gateway+atomic', '前言'],
|
||
['README_EN.md', /^(\d+) reverse engineering skills/m, 'total', '首段技能总数'],
|
||
['README_EN.md', /^## Skill map \((\d+)\)/m, 'total', 'Skill map 标题'],
|
||
['README_EN.md', /^Entry → (\d+) category gateways → (\d+) atomic skills/m, 'gateway+atomic', '前言'],
|
||
['AGENTS.md', /^本仓库是通用逆向工程 AI 技能库（(\d+) 个技能）/m, 'total', '首段技能总数'],
|
||
['AGENTS.md', /^- 原子技能（(\d+)）/m, 'atomic', '原子技能行'],
|
||
['CLAUDE.md', /^通用逆向工程 AI 技能库（(\d+) 个技能）/m, 'total', '仓库性质段'],
|
||
['CLAUDE.md', /OK: (\d+) skills validated/, 'total', '校验输出示例'],
|
||
['CLAUDE.md', /原子技能（type: atomic，(\d+) 个）/m, 'atomic', '架构图'],
|
||
['package.json', /"description":\s*"[^"]*?(\d+) 个技能/, 'total', 'npm 描述'],
|
||
['.claude-plugin/marketplace.json', /"description":\s*"[^"]*?(\d+) 个技能/, 'total', '插件市场描述'],
|
||
];
|
||
|
||
test('各处的技能计数与实际目录一致', () => {
|
||
const c = actualCounts();
|
||
const bad = [];
|
||
for (const [file, re, kind, label] of SITES) {
|
||
const text = readFileSync(file, 'utf8');
|
||
const m = re.exec(text);
|
||
if (!m) { bad.push(`${file}（${label}）：找不到计数，模式可能已失效`); continue; }
|
||
const nums = m.slice(1).map(Number);
|
||
const want = kind === 'total' ? [c.total]
|
||
: kind === 'atomic' ? [c.atomic]
|
||
: [c.gateway, c.atomic];
|
||
nums.forEach((n, i) => {
|
||
if (n !== want[i]) bad.push(`${file}（${label}）：写的是 ${n}，实际 ${want[i]}`);
|
||
});
|
||
}
|
||
assert.deepEqual(bad, [], `计数不同步：\n${bad.join('\n')}`);
|
||
});
|
||
|
||
test('README / README_EN 的网关行引用的技能都存在', () => {
|
||
const missing = [];
|
||
for (const file of ['README.md', 'README_EN.md']) {
|
||
const text = readFileSync(file, 'utf8');
|
||
for (const m of text.matchAll(/`(re-[a-z0-9-]+)`/g)) {
|
||
if (!existsSync(`${SKILLS_DIR}/${m[1]}/SKILL.md`)) missing.push(`${file}: ${m[1]}`);
|
||
}
|
||
}
|
||
assert.deepEqual(missing, [], `引用了不存在的技能：\n${missing.join('\n')}`);
|
||
});
|
||
|
||
test('每个技能都在 README 中至少出现一次', () => {
|
||
const readme = readFileSync('README.md', 'utf8');
|
||
const absent = readdirSync(SKILLS_DIR)
|
||
.filter((d) => d.startsWith('re-') && existsSync(`${SKILLS_DIR}/${d}/SKILL.md`))
|
||
.filter((d) => !readme.includes(d));
|
||
assert.deepEqual(absent, [], `README 未提及：${absent.join(' ')}`);
|
||
});
|
||
|
||
// 测试计数也曾漂移过：CLAUDE.md 写「validate + 103 项测试」，实际已到 168。
|
||
// 技能计数有上面的 SITES 看住，测试计数此前无人看——补上同样的检查：
|
||
// 顶层 test( 的静态计数与 `node --test` 报告的 tests 数一致，故可静态比对。
|
||
test('CLAUDE.md 的测试计数与实际测试数一致', () => {
|
||
let n = 0;
|
||
for (const f of readdirSync('tests')) {
|
||
if (!f.endsWith('.test.mjs')) continue;
|
||
n += (readFileSync(`tests/${f}`, 'utf8').match(/^test\(/gm) || []).length;
|
||
}
|
||
const m = /`npm test` = validate \+ (\d+) 项测试/.exec(readFileSync('CLAUDE.md', 'utf8'));
|
||
assert.ok(m, 'CLAUDE.md 的「检查分层」段应写明测试项数');
|
||
assert.equal(
|
||
Number(m[1]), n,
|
||
`CLAUDE.md 写 ${m[1]} 项，实际 ${n} 项——增删测试后需同步该计数`,
|
||
);
|
||
});
|