einvoice/test/suite/einvoice_edge-cases/test.edge-04.unusual-charsets.ts

import { expect, tap } from '@git.zone/tstest/tapbundle';
import { EInvoice } from '../../../ts/index.js';

tap.test('EDGE-04: Unusual Character Sets - should handle unusual and exotic character encodings', async () => {
  console.log('Testing unusual character sets in e-invoices...\n');
  
  // Test 1: Unicode edge cases with real invoice data
  const testUnicodeEdgeCases = async () => {
    const testCases = [
      {
        name: 'zero-width-characters',
        text: 'Invoice\u200B\u200C\u200D\uFEFFNumber',
        description: 'Zero-width spaces and joiners'
      },
      {
        name: 'right-to-left',
        text: 'مرحبا INV-001 שלום',
        description: 'RTL Arabic and Hebrew mixed with LTR'
      },
      {
        name: 'surrogate-pairs',
        text: '𝐇𝐞𝐥𝐥𝐨 😀 🎉 Invoice',
        description: 'Mathematical bold text and emojis'
      },
      {
        name: 'combining-characters',
        text: 'Ińvȯíçë̃ Nüm̈bër̊',
        description: 'Combining diacritical marks'
      },
      {
        name: 'control-characters',
        text: 'Invoice Test',  // Remove actual control chars as they break XML
        description: 'Control characters (removed for XML safety)'
      },
      {
        name: 'bidi-override',
        text: '\u202Eتسا Invoice 123\u202C',
        description: 'Bidirectional override characters'
      }
    ];
    
    const results = [];
    
    for (const testCase of testCases) {
      try {
        const einvoice = new EInvoice();
        einvoice.issueDate = new Date(2024, 0, 1);
        einvoice.id = testCase.text;
        einvoice.subject = testCase.description;
        
        // Set required fields for EN16931 compliance
        einvoice.from = {
          type: 'company',
          name: 'Test Unicode Company',
          description: testCase.description,
          address: {
            streetName: 'Test Street',
            houseNumber: '1',
            postalCode: '12345',
            city: 'Test City',
            country: 'DE'
          },
          status: 'active',
          foundedDate: { year: 2020, month: 1, day: 1 },
          registrationDetails: {
            vatId: 'DE123456789',
            registrationId: 'HRB 12345',
            registrationName: 'Commercial Register'
          }
        };
        
        einvoice.to = {
          type: 'person',
          name: 'Test',
          surname: 'Customer',
          salutation: 'Mr' as const,
          sex: 'male' as const,
          title: 'Doctor' as const,
          description: 'Test customer',
          address: {
            streetName: 'Customer Street',
            houseNumber: '2',
            postalCode: '54321',
            city: 'Customer City',
            country: 'DE'
          }
        };
        
        // Add test item
        einvoice.items = [{
          position: 1,
          name: `Item with ${testCase.name}`,
          articleNumber: 'ART-001',
          unitType: 'EA',
          unitQuantity: 1,
          unitNetPrice: 100,
          vatPercentage: 19
        }];
        
        // Export to UBL format
        const ublString = await einvoice.toXmlString('ubl');
        
        // Check if special characters are preserved
        const preserved = ublString.includes(testCase.text);
        
        // Try to import it back
        const newInvoice = new EInvoice();
        await newInvoice.fromXmlString(ublString);
        
        const roundTripPreserved = (newInvoice.id === testCase.text || 
                                   newInvoice.invoiceId === testCase.text ||
                                   newInvoice.accountingDocId === testCase.text);
        
        console.log(`Test 1.${testCase.name}:`);
        console.log(`  Unicode preserved in XML: ${preserved ? 'Yes' : 'No'}`);
        console.log(`  Round-trip successful: ${roundTripPreserved ? 'Yes' : 'No'}`);
        
        results.push({ name: testCase.name, preserved, roundTripPreserved });
      } catch (error) {
        console.log(`Test 1.${testCase.name}:`);
        console.log(`  Error: ${error.message}`);
        results.push({ name: testCase.name, preserved: false, roundTripPreserved: false, error: error.message });
      }
    }
    
    return results;
  };

  // Test 2: Various character encodings in invoice content
  const testVariousEncodings = async () => {
    const encodingTests = [
      {
        encoding: 'UTF-8',
        text: 'Übung macht den Meister - äöüß'
      },
      {
        encoding: 'Latin',
        text: 'Ñoño español - ¡Hola!'
      },
      {
        encoding: 'Cyrillic',
        text: 'Счёт-фактура № 2024'
      },
      {
        encoding: 'Greek',
        text: 'Τιμολόγιο: ΜΜΚΔ'
      },
      {
        encoding: 'Chinese',
        text: '發票編號：貳零貳肆'
      }
    ];
    
    const results = [];
    
    for (const test of encodingTests) {
      try {
        const einvoice = new EInvoice();
        einvoice.issueDate = new Date(2024, 0, 1);
        einvoice.id = `ENC-${test.encoding}`;
        einvoice.subject = test.text;
        
        einvoice.from = {
          type: 'company',
          name: test.text,
          description: `Company using ${test.encoding}`,
          address: {
            streetName: 'Test Street',
            houseNumber: '1',
            postalCode: '12345',
            city: 'Test City',
            country: 'DE'
          },
          status: 'active',
          foundedDate: { year: 2020, month: 1, day: 1 },
          registrationDetails: {
            vatId: 'DE123456789',
            registrationId: 'HRB 12345',
            registrationName: 'Commercial Register'
          }
        };
        
        einvoice.to = {
          type: 'person',
          name: 'Test',
          surname: 'Customer',
          salutation: 'Mr' as const,
          sex: 'male' as const,
          title: 'Doctor' as const,
          description: 'Test customer',
          address: {
            streetName: 'Customer Street',
            houseNumber: '2',
            postalCode: '54321',
            city: 'Customer City',
            country: 'DE'
          }
        };
        
        einvoice.items = [{
          position: 1,
          name: test.text,
          articleNumber: 'ENC-001',
          unitType: 'EA',
          unitQuantity: 1,
          unitNetPrice: 100,
          vatPercentage: 19
        }];
        
        // Test both UBL and CII formats
        const ublString = await einvoice.toXmlString('ubl');
        const ciiString = await einvoice.toXmlString('cii');
        
        // Check preservation in both formats
        const ublPreserved = ublString.includes(test.text);
        const ciiPreserved = ciiString.includes(test.text);
        
        // Test round-trip for both formats
        const ublInvoice = new EInvoice();
        await ublInvoice.fromXmlString(ublString);
        
        const ciiInvoice = new EInvoice();
        await ciiInvoice.fromXmlString(ciiString);
        
        const ublRoundTrip = ublInvoice.from?.name?.includes(test.text.substring(0, 10)) || false;
        const ciiRoundTrip = ciiInvoice.from?.name?.includes(test.text.substring(0, 10)) || false;
        
        console.log(`\nTest 2.${test.encoding}:`);
        console.log(`  UBL preserves encoding: ${ublPreserved ? 'Yes' : 'No'}`);
        console.log(`  CII preserves encoding: ${ciiPreserved ? 'Yes' : 'No'}`);
        console.log(`  UBL round-trip: ${ublRoundTrip ? 'Yes' : 'No'}`);
        console.log(`  CII round-trip: ${ciiRoundTrip ? 'Yes' : 'No'}`);
        
        results.push({ 
          encoding: test.encoding, 
          ublPreserved, 
          ciiPreserved, 
          ublRoundTrip, 
          ciiRoundTrip 
        });
      } catch (error) {
        console.log(`\nTest 2.${test.encoding}:`);
        console.log(`  Error: ${error.message}`);
        results.push({ 
          encoding: test.encoding, 
          ublPreserved: false, 
          ciiPreserved: false, 
          ublRoundTrip: false, 
          ciiRoundTrip: false, 
          error: error.message 
        });
      }
    }
    
    return results;
  };

  // Test 3: Extremely unusual characters
  const testExtremelyUnusualChars = async () => {
    const extremeTests = [
      {
        name: 'ancient-scripts',
        text: '𐀀𐀁𐀂 Invoice 𓀀𓀁𓀂',
        description: 'Linear B and Egyptian hieroglyphs'
      },
      {
        name: 'musical-symbols',
        text: '♪♫♪ Invoice ♫♪♫',
        description: 'Musical notation symbols'
      },
      {
        name: 'math-symbols',
        text: '∫∂ Invoice ∆∇',
        description: 'Mathematical operators'
      },
      {
        name: 'private-use',
        text: '\uE000\uE001 Invoice \uE002\uE003',
        description: 'Private use area characters'
      }
    ];
    
    const results = [];
    
    for (const test of extremeTests) {
      try {
        const einvoice = new EInvoice();
        einvoice.id = `EXTREME-${test.name}`;
        einvoice.issueDate = new Date(2024, 0, 1);
        einvoice.subject = test.description;
        
        einvoice.from = {
          type: 'company',
          name: `Company ${test.text}`,
          description: test.description,
          address: {
            streetName: 'Test Street',
            houseNumber: '1',
            postalCode: '12345',
            city: 'Test City',
            country: 'DE'
          },
          status: 'active',
          foundedDate: { year: 2020, month: 1, day: 1 },
          registrationDetails: {
            vatId: 'DE123456789',
            registrationId: 'HRB 12345',
            registrationName: 'Commercial Register'
          }
        };
        
        einvoice.to = {
          type: 'person',
          name: 'Test',
          surname: 'Customer',
          salutation: 'Mr' as const,
          sex: 'male' as const,
          title: 'Doctor' as const,
          description: 'Test customer',
          address: {
            streetName: 'Customer Street',
            houseNumber: '2',
            postalCode: '54321',
            city: 'Customer City',
            country: 'DE'
          }
        };
        
        einvoice.items = [{
          position: 1,
          name: `Product ${test.text}`,
          articleNumber: 'EXT-001',
          unitType: 'EA',
          unitQuantity: 1,
          unitNetPrice: 100,
          vatPercentage: 19
        }];
        
        const xmlString = await einvoice.toXmlString('ubl');
        const preserved = xmlString.includes(test.text);
        
        const newInvoice = new EInvoice();
        await newInvoice.fromXmlString(xmlString);
        
        const roundTrip = newInvoice.from?.name?.includes(test.text) || false;
        
        console.log(`\nTest 3.${test.name}:`);
        console.log(`  Extreme chars preserved: ${preserved ? 'Yes' : 'No'}`);
        console.log(`  Round-trip successful: ${roundTrip ? 'Yes' : 'No'}`);
        
        results.push({ name: test.name, preserved, roundTrip });
      } catch (error) {
        console.log(`\nTest 3.${test.name}:`);
        console.log(`  Error: ${error.message}`);
        results.push({ name: test.name, preserved: false, roundTrip: false, error: error.message });
      }
    }
    
    return results;
  };

  // Test 4: Normalization issues
  const testNormalizationIssues = async () => {
    const normalizationTests = [
      {
        name: 'nfc-nfd',
        nfc: 'é',  // NFC: single character
        nfd: 'é',  // NFD: e + combining acute
        description: 'NFC vs NFD normalization'
      },
      {
        name: 'ligatures',
        text: 'ff Invoice ffi',  // ff and ffi ligatures
        description: 'Unicode ligatures'
      }
    ];
    
    const results = [];
    
    for (const test of normalizationTests) {
      try {
        const einvoice = new EInvoice();
        einvoice.id = `NORM-${test.name}`;
        einvoice.issueDate = new Date(2024, 0, 1);
        einvoice.subject = test.description;
        
        // Use the test text in company name
        const testText = test.text || test.nfc;
        
        einvoice.from = {
          type: 'company',
          name: `Company ${testText}`,
          description: test.description,
          address: {
            streetName: 'Test Street',
            houseNumber: '1',
            postalCode: '12345',
            city: 'Test City',
            country: 'DE'
          },
          status: 'active',
          foundedDate: { year: 2020, month: 1, day: 1 },
          registrationDetails: {
            vatId: 'DE123456789',
            registrationId: 'HRB 12345',
            registrationName: 'Commercial Register'
          }
        };
        
        einvoice.to = {
          type: 'person',
          name: 'Test',
          surname: 'Customer',
          salutation: 'Mr' as const,
          sex: 'male' as const,
          title: 'Doctor' as const,
          description: 'Test customer',
          address: {
            streetName: 'Customer Street',
            houseNumber: '2',
            postalCode: '54321',
            city: 'Customer City',
            country: 'DE'
          }
        };
        
        einvoice.items = [{
          position: 1,
          name: `Product ${testText}`,
          articleNumber: 'NORM-001',
          unitType: 'EA',
          unitQuantity: 1,
          unitNetPrice: 100,
          vatPercentage: 19
        }];
        
        const xmlString = await einvoice.toXmlString('ubl');
        const preserved = xmlString.includes(testText);
        
        const newInvoice = new EInvoice();
        await newInvoice.fromXmlString(xmlString);
        
        const roundTrip = newInvoice.from?.name?.includes(testText) || false;
        
        console.log(`\nTest 4.${test.name}:`);
        console.log(`  Normalization preserved: ${preserved ? 'Yes' : 'No'}`);
        console.log(`  Round-trip successful: ${roundTrip ? 'Yes' : 'No'}`);
        
        results.push({ name: test.name, preserved, roundTrip });
      } catch (error) {
        console.log(`\nTest 4.${test.name}:`);
        console.log(`  Error: ${error.message}`);
        results.push({ name: test.name, preserved: false, roundTrip: false, error: error.message });
      }
    }
    
    return results;
  };
  
  // Run all tests
  const unicodeResults = await testUnicodeEdgeCases();
  const encodingResults = await testVariousEncodings();
  const extremeResults = await testExtremelyUnusualChars();
  const normalizationResults = await testNormalizationIssues();
  
  console.log(`\n=== Unusual Character Sets Test Summary ===`);
  
  // Count successful tests
  const unicodeSuccess = unicodeResults.filter(r => r.roundTripPreserved).length;
  const encodingSuccess = encodingResults.filter(r => r.ublRoundTrip || r.ciiRoundTrip).length;
  const extremeSuccess = extremeResults.filter(r => r.roundTrip).length;
  const normalizationSuccess = normalizationResults.filter(r => r.roundTrip).length;
  
  console.log(`Unicode edge cases: ${unicodeSuccess}/${unicodeResults.length} successful`);
  console.log(`Various encodings: ${encodingSuccess}/${encodingResults.length} successful`);
  console.log(`Extreme characters: ${extremeSuccess}/${extremeResults.length} successful`);
  console.log(`Normalization tests: ${normalizationSuccess}/${normalizationResults.length} successful`);
  
  // Test passes if at least basic Unicode handling works
  const basicUnicodeWorks = unicodeResults.some(r => r.roundTripPreserved);
  const basicEncodingWorks = encodingResults.some(r => r.ublRoundTrip || r.ciiRoundTrip);
  
  expect(basicUnicodeWorks).toBeTrue();
  expect(basicEncodingWorks).toBeTrue();
});

// Run the test
tap.start();