tractatus/tests/unit/ProhibitedTermsScanner.test.js

/**
 * Unit Tests - ProhibitedTermsScanner
 * Tests for proactive content scanning (Framework Phase 1)
 */

const fs = require('fs').promises;
const path = require('path');
const ProhibitedTermsScanner = require('../../scripts/framework-components/ProhibitedTermsScanner');

describe('ProhibitedTermsScanner', () => {
  let scanner;
  const testFilesDir = path.join(__dirname, '../tmp-scanner-test');

  beforeEach(() => {
    scanner = new ProhibitedTermsScanner({
      silent: true,
      basePath: testFilesDir  // Only scan test directory
    });
  });

  afterEach(async () => {
    // Clean up test files
    try {
      await fs.rm(testFilesDir, { recursive: true, force: true });
    } catch (err) {
      // Ignore cleanup errors
    }
  });

  describe('Pattern Detection - inst_017 (Absolute Assurance)', () => {
    test('should detect "guarantee" variations', async () => {
      const testContent = `
        This guarantees safety.
        We guarantee results.
        This is guaranteed to work.
        We are guaranteeing success.
      `;

      // Create test file
      await fs.mkdir(testFilesDir, { recursive: true });
      const testFile = path.join(testFilesDir, 'test.md');
      await fs.writeFile(testFile, testContent);

      const violations = await scanner.scan();
      const inst017Violations = violations.filter(v => v.rule === 'inst_017');

      expect(inst017Violations.length).toBeGreaterThanOrEqual(4);
      expect(inst017Violations.some(v => v.match.toLowerCase().includes('guarantee'))).toBe(true);
    });

    test('should detect "ensures 100%"', async () => {
      const testContent = 'This ensures 100% accuracy.';

      await fs.mkdir(testFilesDir, { recursive: true });
      const testFile = path.join(testFilesDir, 'test.html');
      await fs.writeFile(testFile, testContent);

      const violations = await scanner.scan();
      const inst017Violations = violations.filter(v => v.rule === 'inst_017');

      expect(inst017Violations.length).toBeGreaterThan(0);
      expect(inst017Violations.some(v => v.match.toLowerCase().includes('ensures'))).toBe(true);
    });

    test('should detect "eliminates all"', async () => {
      const testContent = 'This eliminates all bugs.';

      await fs.mkdir(testFilesDir, { recursive: true });
      const testFile = path.join(testFilesDir, 'test.js');
      await fs.writeFile(testFile, testContent);

      const violations = await scanner.scan();

      expect(violations.length).toBeGreaterThan(0);
      expect(violations.some(v => v.match.toLowerCase().includes('eliminates'))).toBe(true);
    });

    test('should detect "never fails"', async () => {
      const testContent = 'This system never fails.';

      await fs.mkdir(testFilesDir, { recursive: true });
      const testFile = path.join(testFilesDir, 'test.md');
      await fs.writeFile(testFile, testContent);

      const violations = await scanner.scan();

      expect(violations.length).toBeGreaterThan(0);
      expect(violations.some(v => v.match.toLowerCase().includes('never'))).toBe(true);
    });
  });

  describe('Pattern Detection - inst_018 (Unverified Claims)', () => {
    test('should detect "production-ready" without context', async () => {
      const testContent = 'This is production-ready software.';

      await fs.mkdir(testFilesDir, { recursive: true });
      const testFile = path.join(testFilesDir, 'test.md');
      await fs.writeFile(testFile, testContent);

      const violations = await scanner.scan();
      const inst018Violations = violations.filter(v => v.rule === 'inst_018');

      expect(inst018Violations.length).toBeGreaterThan(0);
    });

    test('should detect "battle-tested"', async () => {
      const testContent = 'Our battle-tested framework.';

      await fs.mkdir(testFilesDir, { recursive: true });
      const testFile = path.join(testFilesDir, 'test.html');
      await fs.writeFile(testFile, testContent);

      const violations = await scanner.scan();

      expect(violations.some(v => v.match.toLowerCase().includes('battle-tested'))).toBe(true);
    });

    test('should allow "production-ready development tool"', async () => {
      const testContent = 'Tractatus is a production-ready development tool.';

      await fs.mkdir(testFilesDir, { recursive: true });
      const testFile = path.join(testFilesDir, 'test.md');
      await fs.writeFile(testFile, testContent);

      const violations = await scanner.scan();
      const inst018Violations = violations.filter(v =>
        v.rule === 'inst_018' && v.match.toLowerCase().includes('production-ready')
      );

      expect(inst018Violations.length).toBe(0);
    });

    test('should allow "production-ready proof-of-concept"', async () => {
      const testContent = 'This is a production-ready proof-of-concept.';

      await fs.mkdir(testFilesDir, { recursive: true });
      const testFile = path.join(testFilesDir, 'test.md');
      await fs.writeFile(testFile, testContent);

      const violations = await scanner.scan();
      const inst018Violations = violations.filter(v =>
        v.rule === 'inst_018' && v.match.toLowerCase().includes('production-ready')
      );

      expect(inst018Violations.length).toBe(0);
    });
  });

  describe('Context Awareness', () => {
    test('should allow prohibited terms in comments about rules', async () => {
      const testContent = `
        // inst_017: Never use "guarantee"
        // inst_017 prohibits guaranteed language
      `;

      await fs.mkdir(testFilesDir, { recursive: true });
      const testFile = path.join(testFilesDir, 'test.js');
      await fs.writeFile(testFile, testContent);

      const violations = await scanner.scan();

      // Should not flag violations in comments about the rules
      expect(violations.length).toBe(0);
    });

    test('should exclude test files from scanning', async () => {
      // Test files should be excluded by pattern
      const scanner2 = new ProhibitedTermsScanner({ silent: true });

      expect(scanner2.excludePatterns).toContain('**/tests/**/*.test.js');
      expect(scanner2.excludePatterns).toContain('**/tests/**/*.spec.js');
    });

    test('should exclude GOVERNANCE-RULE-LIBRARY.md', async () => {
      expect(scanner.excludePatterns).toContain('**/GOVERNANCE-RULE-LIBRARY.md');
    });

    test('should exclude case studies', async () => {
      expect(scanner.excludePatterns).toContain('**/docs/case-studies/**');
    });
  });

  describe('Suggestions', () => {
    test('should suggest "enforcement" for "guarantee"', () => {
      const suggestions = scanner.patterns.find(p => p.id === 'inst_017').suggestions;

      expect(suggestions['guarantee']).toBe('enforcement');
      expect(suggestions['guarantees']).toBe('enforces');
      expect(suggestions['guaranteed']).toBe('enforced');
    });

    test('should suggest replacements for "ensures 100%"', () => {
      const suggestions = scanner.patterns.find(p => p.id === 'inst_017').suggestions;

      expect(suggestions['ensures 100%']).toBe('helps ensure');
    });

    test('should suggest replacements for inst_018 terms', () => {
      const suggestions = scanner.patterns.find(p => p.id === 'inst_018').suggestions;

      expect(suggestions['production-ready']).toBe('proof-of-concept');
      expect(suggestions['battle-tested']).toBe('in development');
    });

    test('should get suggestion for matched term', () => {
      const patternSet = scanner.patterns.find(p => p.id === 'inst_017');
      const suggestion = scanner.getSuggestion('guarantee', patternSet.suggestions);

      expect(suggestion).toBe('enforcement');
    });

    test('should handle case-insensitive matches', () => {
      const patternSet = scanner.patterns.find(p => p.id === 'inst_017');
      const suggestion = scanner.getSuggestion('GUARANTEE', patternSet.suggestions);

      expect(suggestion).toBe('enforcement');
    });
  });

  describe('Auto-fix Functionality', () => {
    test('should fix simple violations', async () => {
      const testContent = 'This guarantees safety.';

      await fs.mkdir(testFilesDir, { recursive: true });
      const testFile = path.join(testFilesDir, 'test.md');
      await fs.writeFile(testFile, testContent);

      const violations = await scanner.scan();
      const results = await scanner.autoFix(violations);

      expect(results.fixed).toBeGreaterThan(0);

      const fixedContent = await fs.readFile(testFile, 'utf8');
      expect(fixedContent).toContain('enforces');  // guarantees → enforces
      expect(fixedContent).not.toContain('guarantees');
    });

    test('should preserve file structure during fix', async () => {
      const testContent = `Line 1: Normal content
Line 2: This guarantees results
Line 3: More content
Line 4: This guaranteed safety
Line 5: Final line`;

      await fs.mkdir(testFilesDir, { recursive: true });
      const testFile = path.join(testFilesDir, 'test.md');
      await fs.writeFile(testFile, testContent);

      const violations = await scanner.scan();
      await scanner.autoFix(violations);

      const fixedContent = await fs.readFile(testFile, 'utf8');
      const lines = fixedContent.split('\n');

      expect(lines.length).toBe(5);
      expect(lines[0]).toBe('Line 1: Normal content');
      expect(lines[4]).toBe('Line 5: Final line');
    });

    test('should handle multiple violations in same file', async () => {
      const testContent = `
        We guarantee success.
        This is guaranteed to work.
        Our guarantees are strong.
      `;

      await fs.mkdir(testFilesDir, { recursive: true });
      const testFile = path.join(testFilesDir, 'test.md');
      await fs.writeFile(testFile, testContent);

      const violations = await scanner.scan();
      const results = await scanner.autoFix(violations);

      expect(results.fixed).toBeGreaterThan(0);

      const fixedContent = await fs.readFile(testFile, 'utf8');
      expect(fixedContent).not.toContain('guarantee');
    });

    test('should skip violations without clear suggestions', async () => {
      // inst_016 violations (fabricated statistics) require manual review
      const testContent = '95% faster performance.';

      await fs.mkdir(testFilesDir, { recursive: true });
      const testFile = path.join(testFilesDir, 'test.md');
      await fs.writeFile(testFile, testContent);

      const violations = await scanner.scan();
      const inst016Violations = violations.filter(v => v.rule === 'inst_016');

      const results = await scanner.autoFix(inst016Violations);

      // inst_016 requires source citation, can't auto-fix
      expect(results.skipped).toBeGreaterThan(0);
    });
  });

  describe('Formatting', () => {
    test('should format empty violations list', () => {
      const output = scanner.formatViolations([]);

      expect(output).toContain('No prohibited terms found');
    });

    test('should format violations summary', () => {
      const violations = [
        { rule: 'inst_017', file: 'test.md', line: 1, match: 'guarantee', severity: 'HIGH' },
        { rule: 'inst_017', file: 'test.md', line: 2, match: 'guaranteed', severity: 'HIGH' },
        { rule: 'inst_018', file: 'test.html', line: 5, match: 'battle-tested', severity: 'MEDIUM' }
      ];

      const output = scanner.formatViolations(violations, false);

      expect(output).toContain('Found 3 violation');
      expect(output).toContain('inst_017: 2');
      expect(output).toContain('inst_018: 1');
    });

    test('should format detailed violations', () => {
      const violations = [
        {
          rule: 'inst_017',
          file: 'test.md',
          line: 1,
          match: 'guarantee',
          severity: 'HIGH',
          context: 'This guarantees safety',
          suggestion: 'enforcement'
        }
      ];

      const output = scanner.formatViolations(violations, true);

      expect(output).toContain('test.md:1');
      expect(output).toContain('Found: "guarantee"');
      expect(output).toContain('Suggestion: enforcement');
    });
  });

  describe('Utility Functions', () => {
    test('should escape regex special characters', () => {
      const escaped = scanner.escapeRegex('test.file[0]');

      expect(escaped).toBe('test\\.file\\[0\\]');
    });

    test('should check allowed context for inst_017 references', () => {
      const line = 'inst_017 prohibits guaranteed language';
      const result = scanner.isAllowedContext(line, 'guaranteed', 'test.md');

      expect(result).toBe(true);
    });

    test('should check allowed context for case studies', () => {
      const line = 'This guarantees results';
      const result = scanner.isAllowedContext(line, 'guarantees', 'docs/case-studies/example.md');

      expect(result).toBe(true);
    });

    test('should reject prohibited terms in normal content', () => {
      const line = 'This guarantees results';
      const result = scanner.isAllowedContext(line, 'guarantees', 'README.md');

      expect(result).toBe(false);
    });
  });

  describe('File Inclusion/Exclusion', () => {
    test('should include markdown files', () => {
      expect(scanner.includePatterns).toContain('**/*.md');
    });

    test('should include HTML files', () => {
      expect(scanner.includePatterns).toContain('**/*.html');
    });

    test('should include JavaScript files', () => {
      expect(scanner.includePatterns).toContain('**/*.js');
    });

    test('should exclude node_modules', () => {
      expect(scanner.excludePatterns).toContain('**/node_modules/**');
    });

    test('should exclude .git directory', () => {
      expect(scanner.excludePatterns).toContain('**/.git/**');
    });

    test('should exclude .claude directory', () => {
      expect(scanner.excludePatterns).toContain('**/.claude/**');
    });
  });

  describe('Edge Cases', () => {
    test('should handle empty files', async () => {
      await fs.mkdir(testFilesDir, { recursive: true });
      const testFile = path.join(testFilesDir, 'empty.md');
      await fs.writeFile(testFile, '');

      const violations = await scanner.scan();

      // Should not error on empty files
      expect(Array.isArray(violations)).toBe(true);
    });

    test('should handle files with only whitespace', async () => {
      await fs.mkdir(testFilesDir, { recursive: true });
      const testFile = path.join(testFilesDir, 'whitespace.md');
      await fs.writeFile(testFile, '\n\n   \n\t\n');

      const violations = await scanner.scan();

      expect(Array.isArray(violations)).toBe(true);
    });

    test('should handle very long lines', async () => {
      const longLine = 'a'.repeat(10000) + ' guarantee ' + 'b'.repeat(10000);

      await fs.mkdir(testFilesDir, { recursive: true });
      const testFile = path.join(testFilesDir, 'long.md');
      await fs.writeFile(testFile, longLine);

      const violations = await scanner.scan();

      expect(violations.length).toBeGreaterThan(0);
    });

    test('should handle violations at file end without newline', async () => {
      const testContent = 'This guarantees results';

      await fs.mkdir(testFilesDir, { recursive: true });
      const testFile = path.join(testFilesDir, 'test.md');
      await fs.writeFile(testFile, testContent);

      const violations = await scanner.scan();

      expect(violations.length).toBeGreaterThan(0);
    });
  });

  describe('Silent Mode', () => {
    test('should suppress console output in silent mode', async () => {
      const silentScanner = new ProhibitedTermsScanner({
        silent: true,
        basePath: testFilesDir
      });
      const consoleSpy = jest.spyOn(console, 'log').mockImplementation(() => {});

      await silentScanner.scan();

      // In silent mode, should not call console.log for scanning message
      const scanningCalls = consoleSpy.mock.calls.filter(call =>
        call.some(arg => typeof arg === 'string' && arg.includes('Scanning'))
      );
      expect(scanningCalls.length).toBe(0);

      consoleSpy.mockRestore();
    });
  });
});