google-drive-content-adapter/tests/unit/proxy-sitemap.test.js.old

/**
 * Unit Tests: Sitemap Generation Logic
 * 
 * Tests sitemap XML generation functions
 * Tests T028, T029, T030
 */

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

describe('Unit: escapeXml() (T028)', () => {
  
  // Mock XML escape function (will be in proxy.js)
  function escapeXml(str) {
    if (typeof str !== 'string') return '';
    
    return str
      .replace(/&/g, '&')
      .replace(/</g, '&lt;')
      .replace(/>/g, '&gt;')
      .replace(/"/g, '&quot;')
      .replace(/'/g, '&apos;');
  }

  it('T028: should escape < character to &lt;', () => {
    // Given: String with < character
    const input = 'test < value';
    
    // When: Escaping for XML
    const output = escapeXml(input);
    
    // Then: Should escape <
    assert.equal(output, 'test &lt; value', 'Should escape <');
  });

  it('T028: should escape > character to &gt;', () => {
    // Given: String with > character
    const input = 'test > value';
    
    // When: Escaping for XML
    const output = escapeXml(input);
    
    // Then: Should escape >
    assert.equal(output, 'test &gt; value', 'Should escape >');
  });

  it('T028: should escape & character to &amp;', () => {
    // Given: String with & character
    const input = 'test & value';
    
    // When: Escaping for XML
    const output = escapeXml(input);
    
    // Then: Should escape &
    assert.equal(output, 'test &amp; value', 'Should escape &');
  });

  it('T028: should escape " character to &quot;', () => {
    // Given: String with " character
    const input = 'test "value"';
    
    // When: Escaping for XML
    const output = escapeXml(input);
    
    // Then: Should escape "
    assert.equal(output, 'test &quot;value&quot;', 'Should escape "');
  });

  it('T028: should escape \' character to &apos;', () => {
    // Given: String with ' character
    const input = "test 'value'";
    
    // When: Escaping for XML
    const output = escapeXml(input);
    
    // Then: Should escape '
    assert.equal(output, 'test &apos;value&apos;', 'Should escape \'');
  });

  it('T028: should escape multiple special characters in correct order', () => {
    // Given: String with multiple special characters
    const input = '<tag attr="value" other=\'test\'>content & more</tag>';
    
    // When: Escaping for XML
    const output = escapeXml(input);
    
    // Then: Should escape all characters properly
    assert.equal(
      output,
      '&lt;tag attr=&quot;value&quot; other=&apos;test&apos;&gt;content &amp; more&lt;/tag&gt;',
      'Should escape all XML special characters'
    );
  });

  it('T028: should handle strings without special characters', () => {
    // Given: String without special characters
    const input = 'normal text 123';
    
    // When: Escaping for XML
    const output = escapeXml(input);
    
    // Then: Should return unchanged
    assert.equal(output, input, 'Should not modify strings without special chars');
  });

  it('T028: should handle empty string', () => {
    // Given: Empty string
    const input = '';
    
    // When: Escaping for XML
    const output = escapeXml(input);
    
    // Then: Should return empty string
    assert.equal(output, '', 'Should handle empty string');
  });

  it('T028: should handle non-string input gracefully', () => {
    // Given: Non-string inputs
    const inputs = [null, undefined, 123, { foo: 'bar' }];
    
    // When: Escaping each input
    // Then: Should return empty string for non-strings
    inputs.forEach(input => {
      const output = escapeXml(input);
      assert.equal(output, '', `Should return empty string for ${typeof input}`);
    });
  });
});

describe('Unit: formatSitemapEntry() (T029)', () => {
  
  // Mock sitemap entry formatter (will be in proxy.js)
  function formatSitemapEntry(document, baseUrl) {
    function escapeXml(str) {
      return str.replace(/&/g, '&amp;').replace(/</g, '&lt;').replace(/>/g, '&gt;');
    }
    
    const loc = `${baseUrl}/${document.id}`;
    const lastmod = document.modifiedTime;
    
    return `  <url>
    <loc>${escapeXml(loc)}</loc>
    <lastmod>${lastmod}</lastmod>
  </url>`;
  }

  it('T029: should convert DriveDocument to XML url element', () => {
    // Given: DriveDocument metadata
    const document = {
      id: '1BxAA_test123',
      name: 'Test Document',
      modifiedTime: '2026-03-06T10:30:00Z'
    };
    const baseUrl = 'http://localhost:3000';
    
    // When: Formatting sitemap entry
    const xml = formatSitemapEntry(document, baseUrl);
    
    // Then: Should generate valid XML
    assert.ok(xml.includes('<url>'), 'Should contain opening url tag');
    assert.ok(xml.includes('</url>'), 'Should contain closing url tag');
    assert.ok(xml.includes('<loc>'), 'Should contain loc element');
    assert.ok(xml.includes('</loc>'), 'Should contain closing loc tag');
    assert.ok(xml.includes('<lastmod>'), 'Should contain lastmod element');
    assert.ok(xml.includes('</lastmod>'), 'Should contain closing lastmod tag');
  });

  it('T029: should include correct location URL with documentId', () => {
    // Given: DriveDocument metadata
    const document = {
      id: '1BxAA_test123',
      name: 'Test Document',
      modifiedTime: '2026-03-06T10:30:00Z'
    };
    const baseUrl = 'http://localhost:3000';
    
    // When: Formatting sitemap entry
    const xml = formatSitemapEntry(document, baseUrl);
    
    // Then: Location should point to adapter endpoint
    assert.ok(
      xml.includes(`<loc>http://localhost:3000/${document.id}</loc>`),
      'Should include correct location URL'
    );
  });

  it('T029: should include ISO 8601 lastmod timestamp', () => {
    // Given: DriveDocument with modified time
    const document = {
      id: '1BxAA_test123',
      name: 'Test Document',
      modifiedTime: '2026-03-06T10:30:00Z'
    };
    const baseUrl = 'http://localhost:3000';
    
    // When: Formatting sitemap entry
    const xml = formatSitemapEntry(document, baseUrl);
    
    // Then: Should include lastmod with ISO 8601 timestamp
    assert.ok(
      xml.includes('<lastmod>2026-03-06T10:30:00Z</lastmod>'),
      'Should include ISO 8601 lastmod timestamp'
    );
  });

  it('T029: should escape special XML characters in URL', () => {
    // Given: DriveDocument with special characters in ID (edge case)
    const document = {
      id: '1BxAA-test&123',
      name: 'Test Document',
      modifiedTime: '2026-03-06T10:30:00Z'
    };
    const baseUrl = 'http://localhost:3000';
    
    // When: Formatting sitemap entry
    const xml = formatSitemapEntry(document, baseUrl);
    
    // Then: Should escape & in URL
    assert.ok(
      xml.includes('&amp;'),
      'Should escape special XML characters in URL'
    );
  });

  it('T029: should handle different baseUrl formats', () => {
    // Given: Different baseUrl formats
    const document = {
      id: '1BxAA_test',
      name: 'Test',
      modifiedTime: '2026-03-06T10:30:00Z'
    };
    
    const baseUrls = [
      'http://localhost:3000',
      'https://example.com',
      'https://api.example.com/v1'
    ];
    
    // When: Formatting with each baseUrl
    // Then: Should generate correct loc for each
    baseUrls.forEach(baseUrl => {
      const xml = formatSitemapEntry(document, baseUrl);
      assert.ok(
        xml.includes(`<loc>${baseUrl}/${document.id}</loc>`),
        `Should work with baseUrl: ${baseUrl}`
      );
    });
  });
});

describe('Unit: generateSitemap() Structure (T030)', () => {
  
  // Mock sitemap generator structure (will be in proxy.js)
  function buildSitemapXml(documents, baseUrl) {
    function escapeXml(str) {
      return str.replace(/&/g, '&amp;').replace(/</g, '&lt;').replace(/>/g, '&gt;');
    }
    
    let xml = '<?xml version="1.0" encoding="UTF-8"?>\n';
    xml += '<urlset xmlns="http://www.sitemaps.org/schemas/sitemap/0.9">\n';
    
    documents.forEach(doc => {
      const loc = `${baseUrl}/${doc.id}`;
      xml += `  <url>\n`;
      xml += `    <loc>${escapeXml(loc)}</loc>\n`;
      xml += `    <lastmod>${doc.modifiedTime}</lastmod>\n`;
      xml += `  </url>\n`;
    });
    
    xml += '</urlset>';
    
    return xml;
  }

  it('T030: should build complete XML with declaration', () => {
    // Given: Array of documents
    const documents = [
      { id: '1BxAA_doc1', name: 'Doc 1', modifiedTime: '2026-03-06T10:00:00Z' }
    ];
    const baseUrl = 'http://localhost:3000';
    
    // When: Building sitemap XML
    const xml = buildSitemapXml(documents, baseUrl);
    
    // Then: Should start with XML declaration
    assert.ok(
      xml.startsWith('<?xml version="1.0"'),
      'Should start with XML declaration'
    );
  });

  it('T030: should include correct sitemap namespace', () => {
    // Given: Array of documents
    const documents = [
      { id: '1BxAA_doc1', name: 'Doc 1', modifiedTime: '2026-03-06T10:00:00Z' }
    ];
    const baseUrl = 'http://localhost:3000';
    
    // When: Building sitemap XML
    const xml = buildSitemapXml(documents, baseUrl);
    
    // Then: Should include sitemap protocol namespace
    assert.ok(
      xml.includes('<urlset xmlns="http://www.sitemaps.org/schemas/sitemap/0.9">'),
      'Should include correct sitemap namespace'
    );
  });

  it('T030: should include closing urlset tag', () => {
    // Given: Array of documents
    const documents = [
      { id: '1BxAA_doc1', name: 'Doc 1', modifiedTime: '2026-03-06T10:00:00Z' }
    ];
    const baseUrl = 'http://localhost:3000';
    
    // When: Building sitemap XML
    const xml = buildSitemapXml(documents, baseUrl);
    
    // Then: Should end with closing urlset tag
    assert.ok(xml.endsWith('</urlset>'), 'Should end with closing urlset tag');
  });

  it('T030: should include multiple url entries for multiple documents', () => {
    // Given: Multiple documents
    const documents = [
      { id: '1BxAA_doc1', name: 'Doc 1', modifiedTime: '2026-03-06T10:00:00Z' },
      { id: '2CyBB_doc2', name: 'Doc 2', modifiedTime: '2026-03-06T11:00:00Z' },
      { id: '3DzCC_doc3', name: 'Doc 3', modifiedTime: '2026-03-06T12:00:00Z' }
    ];
    const baseUrl = 'http://localhost:3000';
    
    // When: Building sitemap XML
    const xml = buildSitemapXml(documents, baseUrl);
    
    // Then: Should include all documents
    const urlCount = (xml.match(/<url>/g) || []).length;
    assert.equal(urlCount, 3, 'Should include 3 url entries');
    
    // Then: Each document should have its loc
    documents.forEach(doc => {
      assert.ok(
        xml.includes(`<loc>http://localhost:3000/${doc.id}</loc>`),
        `Should include url entry for ${doc.id}`
      );
    });
  });

  it('T030: should handle empty document list', () => {
    // Given: Empty documents array
    const documents = [];
    const baseUrl = 'http://localhost:3000';
    
    // When: Building sitemap XML
    const xml = buildSitemapXml(documents, baseUrl);
    
    // Then: Should still have valid XML structure
    assert.ok(xml.includes('<?xml version'), 'Should have XML declaration');
    assert.ok(xml.includes('<urlset'), 'Should have urlset opening');
    assert.ok(xml.includes('</urlset>'), 'Should have urlset closing');
    
    // Then: Should have no url entries
    const urlCount = (xml.match(/<url>/g) || []).length;
    assert.equal(urlCount, 0, 'Should have no url entries');
  });

  it('T030: should generate valid XML that browsers can parse', () => {
    // Given: Sample documents
    const documents = [
      { id: '1BxAA_test', name: 'Test', modifiedTime: '2026-03-06T10:00:00Z' }
    ];
    const baseUrl = 'http://localhost:3000';
    
    // When: Building sitemap XML
    const xml = buildSitemapXml(documents, baseUrl);
    
    // Then: XML should be well-formed (basic checks)
    // Count opening and closing tags
    const openingUrlset = (xml.match(/<urlset/g) || []).length;
    const closingUrlset = (xml.match(/<\/urlset>/g) || []).length;
    assert.equal(openingUrlset, closingUrlset, 'urlset tags should be balanced');
    
    const openingUrl = (xml.match(/<url>/g) || []).length;
    const closingUrl = (xml.match(/<\/url>/g) || []).length;
    assert.equal(openingUrl, closingUrl, 'url tags should be balanced');
  });
});