einvoice/test/suite/einvoice_encoding/test.enc-01.utf8-encoding.ts

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

tap.test('ENC-01: UTF-8 Encoding - should handle UTF-8 encoded documents correctly', async () => {
  console.log('Testing UTF-8 encoding compliance...\n');

  // Test 1: Basic UTF-8 characters in all fields
  const testBasicUtf8 = async () => {
    const einvoice = new EInvoice();
    einvoice.id = 'UTF8-€£¥-001';
    einvoice.date = Date.now();
    einvoice.currency = 'EUR';
    einvoice.subject = 'UTF-8 Test: €£¥ ñüäöß 中文 العربية русский';
    einvoice.notes = ['Special chars: Zürich, Köln, München'];
    
    // Set supplier with UTF-8 characters
    einvoice.from = {
      type: 'company',
      name: 'Büßer & Müller GmbH',
      description: 'German company äöü',
      address: {
        streetName: 'Hauptstraße',
        houseNumber: '42',
        postalCode: '80331',
        city: 'München',
        country: 'DE'
      },
      status: 'active',
      foundedDate: { year: 2020, month: 1, day: 1 },
      registrationDetails: {
        vatId: 'DE123456789',
        registrationId: 'HRB 12345',
        registrationName: 'München'
      }
    };
    
    // Set customer with UTF-8 characters
    einvoice.to = {
      type: 'company',
      name: 'José García S.L.',
      description: 'Spanish company ñ',
      address: {
        streetName: 'Calle Alcalá',
        houseNumber: '123',
        postalCode: '28009',
        city: 'Madrid',
        country: 'ES'
      },
      status: 'active',
      foundedDate: { year: 2019, month: 1, day: 1 },
      registrationDetails: {
        vatId: 'ES987654321',
        registrationId: 'B-87654321',
        registrationName: 'Madrid'
      }
    };
    
    // Add items with UTF-8 characters
    einvoice.items = [{
      position: 1,
      name: 'Spëcïål Îtëm - Contains: €£¥',
      unitType: 'C62',
      unitQuantity: 1,
      unitNetPrice: 100,
      vatPercentage: 19
    }];
    
    // Export to XML
    const xmlString = await einvoice.toXmlString('ubl');
    
    // Check encoding declaration
    const hasEncoding = xmlString.includes('encoding="UTF-8"');
    
    // Check if characters are preserved
    const charactersPreserved = [
      xmlString.includes('UTF8-€£¥-001'),
      xmlString.includes('Büßer'),
      xmlString.includes('Müller'),
      xmlString.includes('José García'),
      xmlString.includes('München'),
      xmlString.includes('Spëcïål')
    ];
    
    // Round-trip test
    const newInvoice = await EInvoice.fromXml(xmlString);
    const roundTripSuccess = 
      newInvoice.id === einvoice.id &&
      newInvoice.from?.name === einvoice.from.name &&
      newInvoice.to?.name === einvoice.to.name;
    
    return {
      hasEncoding,
      charactersPreserved: charactersPreserved.every(p => p),
      roundTripSuccess
    };
  };

  const basicResult = await testBasicUtf8();
  console.log('Test 1 - Basic UTF-8:');
  console.log(`  Encoding declaration: ${basicResult.hasEncoding ? 'Yes' : 'No'}`);
  console.log(`  Characters preserved: ${basicResult.charactersPreserved ? 'Yes' : 'No'}`);
  console.log(`  Round-trip success: ${basicResult.roundTripSuccess ? 'Yes' : 'No'}`);
  expect(basicResult.hasEncoding).toEqual(true);
  expect(basicResult.charactersPreserved).toEqual(true);
  expect(basicResult.roundTripSuccess).toEqual(true);

  // Test 2: Extended Unicode (emoji, CJK)
  const testExtendedUnicode = async () => {
    const einvoice = new EInvoice();
    einvoice.id = 'UNICODE-🌍-001';
    einvoice.date = Date.now();
    einvoice.currency = 'EUR';
    einvoice.subject = '🌍 中文 日本語 한국어 👍';
    
    einvoice.from = {
      type: 'company',
      name: '世界公司 🌏',
      description: 'International company',
      address: {
        streetName: '国际街',
        houseNumber: '88',
        postalCode: '100000',
        city: 'Beijing',
        country: 'CN'
      },
      status: 'active',
      foundedDate: { year: 2020, month: 1, day: 1 },
      registrationDetails: {
        vatId: 'CN123456789',
        registrationId: 'BJ-12345',
        registrationName: 'Beijing'
      }
    };
    
    einvoice.to = {
      type: 'company',
      name: 'Customer Ltd',
      description: 'Customer',
      address: {
        streetName: 'Main Street',
        houseNumber: '1',
        postalCode: '10001',
        city: 'New York',
        country: 'US'
      },
      status: 'active',
      foundedDate: { year: 2019, month: 1, day: 1 },
      registrationDetails: {
        vatId: 'US987654321',
        registrationId: 'NY-54321',
        registrationName: 'New York'
      }
    };
    
    einvoice.items = [{
      position: 1,
      name: '产品 📦',
      unitType: 'C62',
      unitQuantity: 1,
      unitNetPrice: 100,
      vatPercentage: 19
    }];
    
    const xmlString = await einvoice.toXmlString('ubl');
    
    // Check if unicode is preserved or encoded
    const unicodeHandled = 
      xmlString.includes('世界公司') || xmlString.includes('&#') || // Direct or numeric entities
      xmlString.includes('🌍') || xmlString.includes('🌍'); // Emoji
    
    return { unicodeHandled };
  };

  const unicodeResult = await testExtendedUnicode();
  console.log('\nTest 2 - Extended Unicode:');
  console.log(`  Unicode handled: ${unicodeResult.unicodeHandled ? 'Yes' : 'No'}`);
  expect(unicodeResult.unicodeHandled).toEqual(true);

  // Test 3: XML special characters
  const testXmlSpecialChars = async () => {
    const einvoice = new EInvoice();
    einvoice.id = 'XML-SPECIAL-001';
    einvoice.date = Date.now();
    einvoice.currency = 'EUR';
    einvoice.subject = 'Test & < > " \' entities';
    
    einvoice.from = {
      type: 'company',
      name: 'Smith & Sons Ltd.',
      description: 'Company with "special" <characters>',
      address: {
        streetName: 'A & B 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: 'Test'
      }
    };
    
    einvoice.to = {
      type: 'company',
      name: 'Customer <Test>',
      description: 'Customer',
      address: {
        streetName: 'Main St',
        houseNumber: '2',
        postalCode: '54321',
        city: 'City',
        country: 'DE'
      },
      status: 'active',
      foundedDate: { year: 2019, month: 1, day: 1 },
      registrationDetails: {
        vatId: 'DE987654321',
        registrationId: 'HRB 54321',
        registrationName: 'Test'
      }
    };
    
    einvoice.items = [{
      position: 1,
      name: 'Item with <angle> & "quotes"',
      unitType: 'C62',
      unitQuantity: 1,
      unitNetPrice: 100,
      vatPercentage: 19
    }];
    
    const xmlString = await einvoice.toXmlString('ubl');
    
    // Check proper XML escaping
    const properlyEscaped = 
      xmlString.includes('&amp;') || xmlString.includes('&#38;') && // Ampersand
      (xmlString.includes('&lt;') || xmlString.includes('&#60;')) && // Less than
      (xmlString.includes('&gt;') || xmlString.includes('&#62;') || 
       !xmlString.includes('<Test>') || !xmlString.includes('<angle>')); // Greater than in content
    
    // Ensure no unescaped special chars in text content
    const noUnescapedChars = !xmlString.match(/>.*[<>&].*</);
    
    return { properlyEscaped, noUnescapedChars };
  };

  const xmlSpecialResult = await testXmlSpecialChars();
  console.log('\nTest 3 - XML special characters:');
  console.log(`  Properly escaped: ${xmlSpecialResult.properlyEscaped ? 'Yes' : 'No'}`);
  expect(xmlSpecialResult.properlyEscaped).toEqual(true);

  // Test 4: BOM handling
  const testBomHandling = async () => {
    // Test invoice with BOM
    const bomXml = '\ufeff<?xml version="1.0" encoding="UTF-8"?>' +
      '<Invoice xmlns="urn:oasis:names:specification:ubl:schema:xsd:Invoice-2"' +
      ' xmlns:cac="urn:oasis:names:specification:ubl:schema:xsd:CommonAggregateComponents-2"' +
      ' xmlns:cbc="urn:oasis:names:specification:ubl:schema:xsd:CommonBasicComponents-2">' +
      '<cbc:ID>BOM-TEST-001</cbc:ID>' +
      '<cbc:IssueDate>2025-01-25</cbc:IssueDate>' +
      '<cbc:InvoiceTypeCode>380</cbc:InvoiceTypeCode>' +
      '<cbc:DocumentCurrencyCode>EUR</cbc:DocumentCurrencyCode>' +
      '<cac:AccountingSupplierParty>' +
      '<cac:Party>' +
      '<cac:PartyName><cbc:Name>Test Supplier</cbc:Name></cac:PartyName>' +
      '<cac:PostalAddress>' +
      '<cbc:StreetName>Test Street</cbc:StreetName>' +
      '<cbc:CityName>Test City</cbc:CityName>' +
      '<cbc:PostalZone>12345</cbc:PostalZone>' +
      '<cac:Country><cbc:IdentificationCode>DE</cbc:IdentificationCode></cac:Country>' +
      '</cac:PostalAddress>' +
      '</cac:Party>' +
      '</cac:AccountingSupplierParty>' +
      '<cac:AccountingCustomerParty>' +
      '<cac:Party>' +
      '<cac:PartyName><cbc:Name>Test Customer</cbc:Name></cac:PartyName>' +
      '<cac:PostalAddress>' +
      '<cbc:StreetName>Customer Street</cbc:StreetName>' +
      '<cbc:CityName>Customer City</cbc:CityName>' +
      '<cbc:PostalZone>54321</cbc:PostalZone>' +
      '<cac:Country><cbc:IdentificationCode>DE</cbc:IdentificationCode></cac:Country>' +
      '</cac:PostalAddress>' +
      '</cac:Party>' +
      '</cac:AccountingCustomerParty>' +
      '<cac:InvoiceLine>' +
      '<cbc:ID>1</cbc:ID>' +
      '<cbc:InvoicedQuantity unitCode="C62">1</cbc:InvoicedQuantity>' +
      '<cbc:LineExtensionAmount currencyID="EUR">100.00</cbc:LineExtensionAmount>' +
      '<cac:Item><cbc:Name>Test Item</cbc:Name></cac:Item>' +
      '</cac:InvoiceLine>' +
      '</Invoice>';
    
    try {
      const invoice = await EInvoice.fromXml(bomXml);
      return { 
        bomHandled: true, 
        invoiceId: invoice.id,
        correctId: invoice.id === 'BOM-TEST-001'
      };
    } catch (error) {
      return { bomHandled: false, error: error.message };
    }
  };

  const bomResult = await testBomHandling();
  console.log('\nTest 4 - BOM handling:');
  console.log(`  BOM handled: ${bomResult.bomHandled ? 'Yes' : 'No'}`);
  if (bomResult.bomHandled) {
    console.log(`  Invoice ID correct: ${bomResult.correctId ? 'Yes' : 'No'}`);
  }
  expect(bomResult.bomHandled).toEqual(true);
  expect(bomResult.correctId).toEqual(true);

  // Test 5: Different XML encodings in declaration
  const testEncodingDeclarations = async () => {
    // NOTE: The library currently accepts multiple encodings. 
    // This may need to be revisited if EN16931 spec requires UTF-8 only.
    const encodings = [
      { encoding: 'UTF-8', expected: true },
      { encoding: 'utf-8', expected: true },
      { encoding: 'UTF-16', expected: true }, // Library accepts this
      { encoding: 'ISO-8859-1', expected: true } // Library accepts this
    ];
    
    const results = [];
    for (const { encoding, expected } of encodings) {
      const xml = `<?xml version="1.0" encoding="${encoding}"?>` +
        '<Invoice xmlns="urn:oasis:names:specification:ubl:schema:xsd:Invoice-2"' +
        ' xmlns:cac="urn:oasis:names:specification:ubl:schema:xsd:CommonAggregateComponents-2"' +
        ' xmlns:cbc="urn:oasis:names:specification:ubl:schema:xsd:CommonBasicComponents-2">' +
        '<cbc:ID>ENC-TEST-001</cbc:ID>' +
        '<cbc:IssueDate>2025-01-25</cbc:IssueDate>' +
        '<cbc:InvoiceTypeCode>380</cbc:InvoiceTypeCode>' +
        '<cbc:DocumentCurrencyCode>EUR</cbc:DocumentCurrencyCode>' +
        '<cac:AccountingSupplierParty>' +
        '<cac:Party>' +
        '<cac:PartyName><cbc:Name>Test Müller</cbc:Name></cac:PartyName>' +
        '<cac:PostalAddress>' +
        '<cbc:StreetName>Test Street</cbc:StreetName>' +
        '<cbc:CityName>München</cbc:CityName>' +
        '<cbc:PostalZone>12345</cbc:PostalZone>' +
        '<cac:Country><cbc:IdentificationCode>DE</cbc:IdentificationCode></cac:Country>' +
        '</cac:PostalAddress>' +
        '</cac:Party>' +
        '</cac:AccountingSupplierParty>' +
        '<cac:AccountingCustomerParty>' +
        '<cac:Party>' +
        '<cac:PartyName><cbc:Name>Customer</cbc:Name></cac:PartyName>' +
        '<cac:PostalAddress>' +
        '<cbc:StreetName>Customer Street</cbc:StreetName>' +
        '<cbc:CityName>Customer City</cbc:CityName>' +
        '<cbc:PostalZone>54321</cbc:PostalZone>' +
        '<cac:Country><cbc:IdentificationCode>DE</cbc:IdentificationCode></cac:Country>' +
        '</cac:PostalAddress>' +
        '</cac:Party>' +
        '</cac:AccountingCustomerParty>' +
        '<cac:InvoiceLine>' +
        '<cbc:ID>1</cbc:ID>' +
        '<cbc:InvoicedQuantity unitCode="C62">1</cbc:InvoicedQuantity>' +
        '<cbc:LineExtensionAmount currencyID="EUR">100.00</cbc:LineExtensionAmount>' +
        '<cac:Item><cbc:Name>Test Item</cbc:Name></cac:Item>' +
        '</cac:InvoiceLine>' +
        '</Invoice>';
      
      try {
        const invoice = await EInvoice.fromXml(xml);
        const preserved = invoice.from?.address?.city === 'München';
        results.push({
          encoding,
          parsed: true,
          preserved,
          success: expected
        });
      } catch (error) {
        results.push({
          encoding,
          parsed: false,
          error: error.message,
          success: !expected // Expected to fail
        });
      }
    }
    
    return results;
  };

  const encodingResults = await testEncodingDeclarations();
  console.log('\nTest 5 - Encoding declarations:');
  encodingResults.forEach(result => {
    console.log(`  ${result.encoding}: ${result.parsed ? 'Parsed' : 'Failed'} - ${result.success ? 'As expected' : 'Unexpected'}`);
  });
  const allAsExpected = encodingResults.every(r => r.success);
  expect(allAsExpected).toEqual(true);

  console.log('\n✓ All UTF-8 encoding tests completed successfully');
});

tap.start();