google-drive-content-adapter/tests/contract/sitemap-schema.test.js

/**
 * Contract Tests: /sitemap.xml XML Schema Validation
 * 
 * Tests T020-T023: Verify API contract compliance for sitemap endpoint
 * Reference: specs/001-drive-proxy-adapter/contracts/sitemap-xml-schema.md
 * 
 * @module tests/contract/sitemap-schema
 */

import { describe, it } from 'node:test';
import assert from 'node:assert/strict';

// =============================================================================
// T020: Contract test for /sitemap.xml success response (200 OK)
// =============================================================================

describe('T020: /sitemap.xml Success Response Contract', () => {
  it('should return 200 OK with valid XML structure', async () => {
    // Mock response from sitemap endpoint
    const mockResponse = {
      statusCode: 200,
      headers: {
        'content-type': 'application/xml; charset=utf-8'
      },
      body: `<?xml version="1.0" encoding="UTF-8"?>
<urlset xmlns="http://www.sitemaps.org/schemas/sitemap/0.9">
  <url>
    <loc>http://localhost:3000/documents/abc123</loc>
    <lastmod>2024-03-01</lastmod>
  </url>
</urlset>`
    };

    // Verify status code
    assert.equal(mockResponse.statusCode, 200, 'Status code must be 200');

    // Verify Content-Type header
    assert.equal(
      mockResponse.headers['content-type'],
      'application/xml; charset=utf-8',
      'Content-Type must be application/xml; charset=utf-8'
    );

    // Verify XML structure
    assert.match(mockResponse.body, /^<\?xml version="1\.0" encoding="UTF-8"\?>/, 'Must have XML declaration');
    assert.match(mockResponse.body, /<urlset xmlns="http:\/\/www\.sitemaps\.org\/schemas\/sitemap\/0\.9">/, 'Must have urlset with correct namespace');
    assert.match(mockResponse.body, /<\/urlset>$/, 'Must close urlset tag');

    // Verify URL entry structure
    assert.match(mockResponse.body, /<url>/, 'Must contain url entries');
    assert.match(mockResponse.body, /<loc>.*<\/loc>/, 'Each url must have loc element');
    assert.match(mockResponse.body, /<lastmod>.*<\/lastmod>/, 'Each url should have lastmod element');
  });

  it('should return valid XML with RESTful URL format', async () => {
    const mockResponse = {
      statusCode: 200,
      body: `<?xml version="1.0" encoding="UTF-8"?>
<urlset xmlns="http://www.sitemaps.org/schemas/sitemap/0.9">
  <url>
    <loc>http://localhost:3000/documents/abc123</loc>
  </url>
</urlset>`
    };

    // Verify RESTful URL pattern: /documents/{documentId}
    assert.match(
      mockResponse.body,
      /<loc>http:\/\/[^<]+\/documents\/[^<]+<\/loc>/,
      'URLs must follow RESTful format /documents/{documentId}'
    );
  });
});

// =============================================================================
// T021: Contract test for /sitemap.xml with empty Drive (0 documents)
// =============================================================================

describe('T021: /sitemap.xml Empty Drive Response Contract', () => {
  it('should return valid XML with empty urlset when no documents exist', async () => {
    const mockResponse = {
      statusCode: 200,
      headers: {
        'content-type': 'application/xml; charset=utf-8'
      },
      body: `<?xml version="1.0" encoding="UTF-8"?>
<urlset xmlns="http://www.sitemaps.org/schemas/sitemap/0.9">
</urlset>`
    };

    // Verify status code
    assert.equal(mockResponse.statusCode, 200, 'Status code must be 200 even for empty Drive');

    // Verify empty urlset is valid XML
    assert.match(mockResponse.body, /<urlset xmlns="http:\/\/www\.sitemaps\.org\/schemas\/sitemap\/0\.9">/, 'Must have urlset with namespace');
    assert.match(mockResponse.body, /<\/urlset>/, 'Must close urlset tag');
    
    // Verify no url entries
    assert.doesNotMatch(mockResponse.body, /<url>/, 'Should not contain any url entries');
  });
});

// =============================================================================
// T022: Contract test for XML special character escaping
// =============================================================================

describe('T022: XML Special Character Escaping Contract', () => {
  it('should properly escape XML special characters in URLs', async () => {
    // Document IDs can contain special characters that need escaping in XML
    const mockResponse = {
      statusCode: 200,
      body: `<?xml version="1.0" encoding="UTF-8"?>
<urlset xmlns="http://www.sitemaps.org/schemas/sitemap/0.9">
  <url>
    <loc>http://localhost:3000/documents/test&amp;doc</loc>
  </url>
  <url>
    <loc>http://localhost:3000/documents/doc&lt;123</loc>
  </url>
  <url>
    <loc>http://localhost:3000/documents/doc&gt;456</loc>
  </url>
  <url>
    <loc>http://localhost:3000/documents/doc&quot;test</loc>
  </url>
  <url>
    <loc>http://localhost:3000/documents/doc&apos;xyz</loc>
  </url>
</urlset>`
    };

    // Verify special characters are escaped
    assert.match(mockResponse.body, /&amp;/, 'Ampersand (&) must be escaped as &amp;');
    assert.match(mockResponse.body, /&lt;/, 'Less than (<) must be escaped as &lt;');
    assert.match(mockResponse.body, /&gt;/, 'Greater than (>) must be escaped as &gt;');
    assert.match(mockResponse.body, /&quot;/, 'Double quote (") must be escaped as &quot;');
    assert.match(mockResponse.body, /&apos;/, 'Single quote (\') must be escaped as &apos;');

    // Verify unescaped special characters are NOT present in content
    const locContent = mockResponse.body.match(/<loc>(.*?)<\/loc>/g);
    assert.ok(locContent, 'Must have loc elements');
    
    locContent.forEach(loc => {
      const content = loc.replace(/<\/?loc>/g, '');
      const afterProtocol = content.split('://')[1] || '';
      
      // Only check the path/query part, not the protocol separator
      if (afterProtocol.includes('/')) {
        const pathPart = afterProtocol.substring(afterProtocol.indexOf('/'));
        assert.doesNotMatch(pathPart, /[&<>"'](?!amp;|lt;|gt;|quot;|apos;)/, 'Unescaped special chars must not appear in XML content');
      }
    });
  });
});

// =============================================================================
// T023: Contract test for lastmod date format validation
// =============================================================================

describe('T023: lastmod Date Format Contract', () => {
  it('should format lastmod as ISO 8601 date (YYYY-MM-DD)', async () => {
    const mockResponse = {
      statusCode: 200,
      body: `<?xml version="1.0" encoding="UTF-8"?>
<urlset xmlns="http://www.sitemaps.org/schemas/sitemap/0.9">
  <url>
    <loc>http://localhost:3000/documents/doc1</loc>
    <lastmod>2024-03-01</lastmod>
  </url>
  <url>
    <loc>http://localhost:3000/documents/doc2</loc>
    <lastmod>2024-12-31</lastmod>
  </url>
</urlset>`
    };

    // Extract lastmod values
    const lastmodMatches = mockResponse.body.match(/<lastmod>(.*?)<\/lastmod>/g);
    assert.ok(lastmodMatches, 'Must have lastmod elements');
    assert.ok(lastmodMatches.length > 0, 'Must have at least one lastmod element');

    // Verify each lastmod follows ISO 8601 date format (YYYY-MM-DD)
    lastmodMatches.forEach(lastmodTag => {
      const dateValue = lastmodTag.match(/<lastmod>(.*?)<\/lastmod>/)[1];
      
      // Check format: YYYY-MM-DD
      assert.match(dateValue, /^\d{4}-\d{2}-\d{2}$/, 'lastmod must be in YYYY-MM-DD format');
      
      // Verify it's a valid date
      const date = new Date(dateValue);
      assert.ok(!isNaN(date.getTime()), 'lastmod must be a valid date');
      
      // Verify date components
      const [year, month, day] = dateValue.split('-').map(Number);
      assert.ok(year >= 1000 && year <= 9999, 'Year must be 4 digits');
      assert.ok(month >= 1 && month <= 12, 'Month must be 01-12');
      assert.ok(day >= 1 && day <= 31, 'Day must be 01-31');
    });
  });

  it('should accept full ISO 8601 datetime format if provided', async () => {
    // Sitemap protocol also accepts full datetime with timezone
    const mockResponse = {
      statusCode: 200,
      body: `<?xml version="1.0" encoding="UTF-8"?>
<urlset xmlns="http://www.sitemaps.org/schemas/sitemap/0.9">
  <url>
    <loc>http://localhost:3000/documents/doc1</loc>
    <lastmod>2024-03-01T10:30:00+00:00</lastmod>
  </url>
</urlset>`
    };

    const lastmodMatch = mockResponse.body.match(/<lastmod>(.*?)<\/lastmod>/);
    assert.ok(lastmodMatch, 'Must have lastmod element');
    
    const dateValue = lastmodMatch[1];
    
    // Accept either YYYY-MM-DD or full ISO 8601 with timezone
    const isValidFormat = /^\d{4}-\d{2}-\d{2}(T\d{2}:\d{2}:\d{2}[+-]\d{2}:\d{2})?$/.test(dateValue);
    assert.ok(isValidFormat, 'lastmod must be valid ISO 8601 format');
    
    // Verify it's a valid date
    const date = new Date(dateValue);
    assert.ok(!isNaN(date.getTime()), 'lastmod must be a valid datetime');
  });
});