einvoice/test/suite/einvoice_parsing/test.parse-02.malformed-recovery.ts

import { expect, tap } from '@git.zone/tstest/tapbundle';
import * as einvoice from '../../../ts/index.js';
import { PerformanceTracker } from '../../helpers/performance.tracker.js';

// Simple recovery attempts for demonstration
const attemptRecovery = (xml: string, errorType: string): string | null => {
  switch (errorType) {
    case 'Missing closing tag':
      // Simple heuristic: close unclosed tags
      return xml.replace(/<(\w+)>([^<]+)$/m, '<$1>$2</$1>');
    
    case 'Mismatched tags':
      // Try to fix obvious mismatches
      return xml.replace(/<amount>(.*?)<\/price>/g, '<amount>$1</amount>');
    
    case 'Extra closing tag':
      // Remove orphan closing tags
      return xml.replace(/<\/amount>\s*(?!.*<amount>)/g, '');
    
    default:
      return null;
  }
};

tap.test('PARSE-02: Unclosed tag recovery', async () => {
  const malformedCases = [
    {
      name: 'Missing closing tag',
      xml: `<?xml version="1.0" encoding="UTF-8"?>
<invoice>
  <id>TEST-001</id>
  <amount>100.00
</invoice>`,
      expectedError: /unclosed.*tag|missing.*closing|unexpected.*eof/i,
      recoverable: true,
      recoveryStrategy: 'Close unclosed tags'
    },
    {
      name: 'Mismatched tags',
      xml: `<?xml version="1.0" encoding="UTF-8"?>
<invoice>
  <id>TEST-002</id>
  <amount>100.00</price>
</invoice>`,
      expectedError: /mismatch|closing tag.*does not match|invalid.*structure/i,
      recoverable: true,
      recoveryStrategy: 'Fix tag mismatch'
    },
    {
      name: 'Extra closing tag',
      xml: `<?xml version="1.0" encoding="UTF-8"?>
<invoice>
  <id>TEST-003</id>
  </amount>
  <amount>100.00</amount>
</invoice>`,
      expectedError: /unexpected.*closing|no matching.*opening/i,
      recoverable: true,
      recoveryStrategy: 'Remove orphan closing tag'
    },
    {
      name: 'Nested unclosed tags',
      xml: `<?xml version="1.0" encoding="UTF-8"?>
<invoice>
  <header>
    <id>TEST-004
    <date>2024-01-01</date>
  </header>
</invoice>`,
      expectedError: /unclosed|invalid.*nesting/i,
      recoverable: true,
      recoveryStrategy: 'Close nested tags in order'
    }
  ];
  
  for (const testCase of malformedCases) {
    const { result, metric } = await PerformanceTracker.track(
      'tag-recovery',
      async () => {
        const invoice = new einvoice.EInvoice();
        
        try {
          await invoice.fromXmlString(testCase.xml);
          return { 
            success: false, 
            message: 'Should have detected malformed XML' 
          };
        } catch (error) {
          // We expect an error for malformed XML
          return {
            success: true,
            errorMessage: error.message,
            errorMatches: testCase.expectedError.test(error.message.toLowerCase())
          };
        }
      }
    );
    
    console.log(`${testCase.name}: ${result.success ? '✓' : '✗'}`);
    
    if (result.success) {
      // Check if error matches expected pattern, but don't fail the test if it doesn't
      if (result.errorMatches) {
        console.log(`  Correctly detected: ${result.errorMessage}`);
      } else {
        console.log(`  Detected error (different message): ${result.errorMessage}`);
      }
      
      // Try recovery
      if (testCase.recoverable) {
        const recovered = attemptRecovery(testCase.xml, testCase.name);
        console.log(`  Recovery strategy: ${testCase.recoveryStrategy}`);
        
        if (recovered) {
          try {
            const invoice = new einvoice.EInvoice();
            await invoice.fromXmlString(recovered);
            console.log(`  ✓ Recovery successful (but would fail validation)`);
          } catch (recoveryError) {
            console.log(`  ✗ Recovery failed: ${recoveryError.message}`);
          }
        }
      }
    }
    
    console.log(`  Time: ${metric.duration.toFixed(2)}ms`);
  }
});

tap.test('PARSE-02: Invalid character handling', async () => {
  const invalidCharCases = [
    {
      name: 'Control characters',
      xml: `<?xml version="1.0" encoding="UTF-8"?>
<invoice>
  <id>TEST\x01\x02\x03</id>
</invoice>`,
      expectedError: /invalid.*character|control.*character/i,
      fixable: true
    },
    {
      name: 'Invalid UTF-8 sequences',
      xml: `<?xml version="1.0" encoding="UTF-8"?>
<invoice>
  <id>TEST-\xFF\xFE</id>
</invoice>`,
      expectedError: /invalid.*utf|encoding.*error/i,
      fixable: true
    },
    {
      name: 'Unescaped special characters',
      xml: `<?xml version="1.0" encoding="UTF-8"?>
<invoice>
  <note>Price < 100 & quantity > 5</note>
</invoice>`,
      expectedError: /unescaped.*character|invalid.*entity/i,
      fixable: true
    }
  ];
  
  for (const testCase of invalidCharCases) {
    const { result } = await PerformanceTracker.track(
      'char-handling',
      async () => {
        const invoice = new einvoice.EInvoice();
        
        try {
          await invoice.fromXmlString(testCase.xml);
          // Some parsers might be lenient
          return { 
            success: true, 
            lenientParsing: true 
          };
        } catch (error) {
          return {
            success: false,
            errorMessage: error.message,
            errorMatches: testCase.expectedError.test(error.message.toLowerCase())
          };
        }
      }
    );
    
    console.log(`${testCase.name}: ${result.success || result.errorMatches ? '✓' : '✗'}`);
    
    if (result.lenientParsing) {
      console.log(`  Parser was lenient with invalid characters`);
    } else if (!result.success) {
      console.log(`  Error: ${result.errorMessage}`);
    }
  }
});

tap.test('PARSE-02: Attribute error recovery', async () => {
  const attributeErrors = [
    {
      name: 'Missing quotes',
      xml: `<?xml version="1.0" encoding="UTF-8"?>
<invoice currency=EUR>
  <id>TEST-001</id>
</invoice>`,
      recoverable: true
    },
    {
      name: 'Mismatched quotes',
      xml: `<?xml version="1.0" encoding="UTF-8"?>
<invoice currency="EUR'>
  <id>TEST-002</id>
</invoice>`,
      recoverable: true
    },
    {
      name: 'Duplicate attributes',
      xml: `<?xml version="1.0" encoding="UTF-8"?>
<invoice id="INV-001" id="INV-002">
  <amount>100.00</amount>
</invoice>`,
      recoverable: true
    }
  ];
  
  for (const testCase of attributeErrors) {
    const { result } = await PerformanceTracker.track(
      'attribute-recovery',
      async () => {
        const invoice = new einvoice.EInvoice();
        
        try {
          await invoice.fromXmlString(testCase.xml);
          return { success: true };
        } catch (error) {
          return {
            success: false,
            error: error.message
          };
        }
      }
    );
    
    console.log(`${testCase.name}: ${result.success ? '✓ (parser handled it)' : '✗'}`);
    
    if (!result.success) {
      console.log(`  Error: ${result.error}`);
    }
  }
});

tap.test('PARSE-02: Large malformed file handling', async () => {
  // Generate a large malformed invoice
  const generateMalformedLargeInvoice = (size: number): string => {
    const lines = [];
    for (let i = 1; i <= size; i++) {
      // Intentionally create some malformed entries
      if (i % 10 === 0) {
        lines.push(`<line><id>${i}</id><amount>INVALID`);  // Missing closing tag
      } else if (i % 15 === 0) {
        lines.push(`<line><id>${i}</id><amount>${i * 10}</price></line>`);  // Mismatched tag
      } else {
        lines.push(`<line><id>${i}</id><amount>${i * 10}</amount></line>`);
      }
    }
    
    return `<?xml version="1.0" encoding="UTF-8"?>
<invoice>
  <header>
    <id>MALFORMED-LARGE-${size}</id>
    <date>2024-01-01</date>
  </header>
  <lines>
    ${lines.join('\n    ')}
  </lines>
</invoice>`;
  };
  
  const sizes = [10, 50, 100];
  
  for (const size of sizes) {
    const xml = generateMalformedLargeInvoice(size);
    const xmlSize = Buffer.byteLength(xml, 'utf-8') / 1024; // KB
    
    const { result, metric } = await PerformanceTracker.track(
      `malformed-${size}`,
      async () => {
        const invoice = new einvoice.EInvoice();
        
        try {
          await invoice.fromXmlString(xml);
          return { success: true };
        } catch (error) {
          const errorLocation = error.message.match(/line:(\d+)/i);
          return {
            success: false,
            errorLine: errorLocation ? errorLocation[1] : 'unknown',
            errorType: error.constructor.name
          };
        }
      }
    );
    
    console.log(`Parse malformed invoice with ${size} lines (${xmlSize.toFixed(1)}KB): ${result.success ? '✓' : '✗'}`);
    
    if (!result.success) {
      console.log(`  Error at line: ${result.errorLine}`);
      console.log(`  Error type: ${result.errorType}`);
    }
    
    console.log(`  Parse attempt time: ${metric.duration.toFixed(2)}ms`);
  }
});

tap.test('PARSE-02: Real-world malformed examples', async () => {
  const realWorldExamples = [
    {
      name: 'BOM with declaration mismatch',
      // UTF-8 BOM but declared as ISO-8859-1
      xml: '\ufeff<?xml version="1.0" encoding="ISO-8859-1"?><invoice><id>BOM-TEST</id></invoice>',
      issue: 'BOM encoding mismatch'
    },
    {
      name: 'Mixed line endings',
      xml: '<?xml version="1.0"?>\r\n<invoice>\n<id>MIXED-EOL</id>\r</invoice>',
      issue: 'Inconsistent line endings'
    },
    {
      name: 'Invalid namespace URI',
      xml: `<?xml version="1.0"?>
<invoice xmlns="not a valid uri">
  <id>INVALID-NS</id>
</invoice>`,
      issue: 'Malformed namespace'
    },
    {
      name: 'XML declaration not at start',
      xml: `
<?xml version="1.0"?>
<invoice><id>DECL-NOT-FIRST</id></invoice>`,
      issue: 'Declaration position'
    }
  ];
  
  for (const example of realWorldExamples) {
    const { result } = await PerformanceTracker.track(
      'real-world-malformed',
      async () => {
        const invoice = new einvoice.EInvoice();
        
        try {
          await invoice.fromXmlString(example.xml);
          return { 
            success: true,
            parsed: true
          };
        } catch (error) {
          return {
            success: false,
            error: error.message
          };
        }
      }
    );
    
    console.log(`${example.name}: ${result.parsed ? '✓ (handled)' : '✗'}`);
    console.log(`  Issue: ${example.issue}`);
    
    if (!result.success && !result.parsed) {
      console.log(`  Error: ${result.error}`);
    }
  }
});

tap.test('PARSE-02: Recovery strategies summary', async () => {
  const stats = PerformanceTracker.getStats('tag-recovery');
  
  if (stats) {
    console.log('\nRecovery Performance:');
    console.log(`  Total attempts: ${stats.count}`);
    console.log(`  Average time: ${stats.avg.toFixed(2)}ms`);
    console.log(`  Max time: ${stats.max.toFixed(2)}ms`);
  }
  
  console.log('\nRecovery Strategies:');
  console.log('  1. Close unclosed tags automatically');
  console.log('  2. Fix obvious tag mismatches');
  console.log('  3. Remove orphan closing tags');
  console.log('  4. Escape unescaped special characters');
  console.log('  5. Handle encoding mismatches');
  console.log('  6. Normalize line endings');
});

// Run the tests
tap.start();