Fixed error message logging, added detailed diagnostics.

2026-06-17 07:43:13 +10:00
parent fdd1d1131d
commit a5145c6902
3 changed files with 394 additions and 10 deletions
@@ -0,0 +1,148 @@
+import { describe, test, expect } from 'bun:test';
+import { analyzeTocPage } from '../src/services/analyzer.ts';
+
+const SCOPE = '#list-chapter .chapter-archive-grid';
+const PATTERN = 'https://novelbin\\.com/b/the-mech-touch/.*';
+
+function makeHtml(opts: {
+  title?: string;
+  body?: string;
+  links?: string[];
+  extraHead?: string;
+}) {
+  const links = (opts.links ?? [])
+    .map(href => `<a href="${href}">link</a>`)
+    .join('');
+  return `<html>
+    <head>
+      <title>${opts.title ?? 'Test Page'}</title>
+      ${opts.extraHead ?? ''}
+    </head>
+    <body>${opts.body ?? ''}${links}</body>
+  </html>`;
+}
+
+describe('analyzer — Cloudflare detection', () => {
+  test('flags "Just a moment" title as Cloudflare', () => {
+    const html = makeHtml({ title: 'Just a moment...' });
+    const d = analyzeTocPage(html, SCOPE, PATTERN);
+    const cf = d.findings.find(f => f.check === 'Cloudflare challenge')!;
+    expect(cf.passed).toBe(false);
+    expect(d.likelyCause).toContain('Cloudflare');
+  });
+
+  test('flags #challenge-form element as Cloudflare', () => {
+    const html = makeHtml({ body: '<form id="challenge-form"></form>' });
+    const d = analyzeTocPage(html, SCOPE, PATTERN);
+    const cf = d.findings.find(f => f.check === 'Cloudflare challenge')!;
+    expect(cf.passed).toBe(false);
+  });
+
+  test('does not flag a clean page as Cloudflare', () => {
+    const html = makeHtml({ title: 'The Mech Touch — Chapters', body: '<div id="list-chapter"><div class="chapter-archive-grid"></div></div>' });
+    const d = analyzeTocPage(html, SCOPE, PATTERN);
+    const cf = d.findings.find(f => f.check === 'Cloudflare challenge')!;
+    expect(cf.passed).toBe(true);
+  });
+});
+
+describe('analyzer — security block detection', () => {
+  test('flags "you have been blocked" body text', () => {
+    const html = makeHtml({ body: '<p>You have been blocked from accessing this site.</p>' });
+    const d = analyzeTocPage(html, SCOPE, PATTERN);
+    const check = d.findings.find(f => f.check === 'Security / bot block')!;
+    expect(check.passed).toBe(false);
+    expect(d.likelyCause).toContain('Security block');
+  });
+});
+
+describe('analyzer — CAPTCHA detection', () => {
+  test('flags g-recaptcha div', () => {
+    const html = makeHtml({ body: '<div class="g-recaptcha" data-sitekey="abc"></div>' });
+    const d = analyzeTocPage(html, SCOPE, PATTERN);
+    const check = d.findings.find(f => f.check === 'CAPTCHA')!;
+    expect(check.passed).toBe(false);
+  });
+});
+
+describe('analyzer — login wall detection', () => {
+  test('flags password input', () => {
+    const html = makeHtml({ body: '<form><input type="password"/></form>' });
+    const d = analyzeTocPage(html, SCOPE, PATTERN);
+    const check = d.findings.find(f => f.check === 'Login wall')!;
+    expect(check.passed).toBe(false);
+    expect(d.likelyCause).toContain('Login');
+  });
+});
+
+describe('analyzer — scope selector breakdown', () => {
+  test('reports root missing when neither segment exists', () => {
+    const html = makeHtml({ body: '<div id="something-else"></div>' });
+    const d = analyzeTocPage(html, SCOPE, PATTERN);
+    const check = d.findings.find(f => f.check === 'TOC scope selector')!;
+    expect(check.passed).toBe(false);
+    expect(check.detail).toContain('#list-chapter');
+    expect(check.detail).toContain('not found anywhere');
+    expect(d.likelyCause).toContain('redesigned');
+  });
+
+  test('reports inner selector missing when root exists but child does not', () => {
+    const html = makeHtml({ body: '<div id="list-chapter"><ul><li>no grid here</li></ul></div>' });
+    const d = analyzeTocPage(html, SCOPE, PATTERN);
+    const check = d.findings.find(f => f.check === 'TOC scope selector')!;
+    expect(check.passed).toBe(false);
+    expect(check.detail).toContain('#list-chapter');
+    expect(check.detail).toContain('.chapter-archive-grid');
+    expect(d.likelyCause).toContain('Inner selector');
+  });
+
+  test('notes when inner selector exists elsewhere on the page (wrong hierarchy)', () => {
+    const html = makeHtml({
+      body: '<div id="list-chapter"></div><div class="chapter-archive-grid"></div>',
+    });
+    const d = analyzeTocPage(html, SCOPE, PATTERN);
+    const check = d.findings.find(f => f.check === 'TOC scope selector')!;
+    expect(check.passed).toBe(false);
+    expect(check.detail).toContain('exists elsewhere');
+  });
+
+  test('passes when full scope matches', () => {
+    const html = makeHtml({
+      body: '<div id="list-chapter"><div class="chapter-archive-grid"></div></div>',
+    });
+    const d = analyzeTocPage(html, SCOPE, PATTERN);
+    const check = d.findings.find(f => f.check === 'TOC scope selector')!;
+    expect(check.passed).toBe(true);
+  });
+});
+
+describe('analyzer — link pattern check', () => {
+  test('finds matching links outside the scope when scope is missing', () => {
+    const links = [
+      'https://novelbin.com/b/the-mech-touch/chapter-1',
+      'https://novelbin.com/b/the-mech-touch/chapter-2',
+    ];
+    const html = makeHtml({ body: '<div id="other">' + links.map(l => `<a href="${l}">ch</a>`).join('') + '</div>' });
+    const d = analyzeTocPage(html, SCOPE, PATTERN);
+    const check = d.findings.find(f => f.check === 'Link pattern (anywhere)')!;
+    expect(check.passed).toBe(true);
+    expect(check.detail).toContain('2 link');
+    expect(d.likelyCause).toContain('scope selector');
+  });
+
+  test('reports no matching links and samples what is there', () => {
+    const html = makeHtml({ links: ['https://other.com/a', 'https://other.com/b'] });
+    const d = analyzeTocPage(html, SCOPE, PATTERN);
+    const check = d.findings.find(f => f.check === 'Link pattern (anywhere)')!;
+    expect(check.passed).toBe(false);
+    expect(check.detail).toContain('other.com');
+  });
+});
+
+describe('analyzer — actualUrl', () => {
+  test('includes the landed URL in the diagnosis', () => {
+    const html = makeHtml({});
+    const d = analyzeTocPage(html, SCOPE, PATTERN, 'https://novelbin.com/redirected');
+    expect(d.actualUrl).toBe('https://novelbin.com/redirected');
+  });
+});