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

import { tap } from '@git.zone/tstest/tapbundle';
import { EInvoice } from '../../../ts/index.js';
import { PerformanceTracker } from '../performance.tracker.js';

tap.test('EDGE-04: Unusual Character Sets - should handle unusual and exotic character encodings', async () => {
  // Test 1: Unicode edge cases with real invoice data
  await PerformanceTracker.track('unicode-edge-cases', 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'
      }
    ];
    
    for (const testCase of testCases) {
      const einvoice = new EInvoice();
      einvoice.issueDate = new Date(2024, 0, 1);
      einvoice.invoiceId = testCase.text;
      einvoice.subject = testCase.description;
      
      // Set required fields
      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
      }];
      
      try {
        // Export to UBL format
        const ublString = await einvoice.toXmlString('ubl');
        
        // Check if special characters are preserved
        const preserved = ublString.includes(testCase.text);
        console.log(`Unicode test ${testCase.name}: ${preserved ? 'preserved' : 'encoded'}`);
        
        // Try to import it back
        const newInvoice = new EInvoice();
        await newInvoice.fromXmlString(ublString);
        
        const roundTripPreserved = newInvoice.invoiceId === testCase.text;
        console.log(`Unicode test ${testCase.name} round-trip: ${roundTripPreserved ? 'success' : 'modified'}`);
      } catch (error) {
        console.log(`Unicode test ${testCase.name} failed: ${error.message}`);
      }
    }
  });

  // Test 2: Various character encodings in invoice content
  await PerformanceTracker.track('various-character-encodings', 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'
      }
    ];
    
    for (const test of encodingTests) {
      const einvoice = new EInvoice();
      einvoice.issueDate = new Date(2024, 0, 1);
      einvoice.invoiceId = `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.text,
          country: 'DE'
        },
        status: 'active',
        foundedDate: { year: 2020, month: 1, day: 1 },
        registrationDetails: {
          vatId: 'DE123456789',
          registrationId: 'HRB 12345',
          registrationName: test.text
        }
      };
      
      einvoice.to = {
        type: 'company',
        name: 'Customer Inc',
        description: 'Test customer',
        address: {
          streetName: 'Customer Street',
          houseNumber: '2',
          postalCode: '54321',
          city: 'Customer City',
          country: 'DE'
        },
        status: 'active',
        foundedDate: { year: 2019, month: 1, day: 1 },
        registrationDetails: {
          vatId: 'DE987654321',
          registrationId: 'HRB 54321',
          registrationName: 'Commercial Register'
        }
      };
      
      einvoice.items = [{
        position: 1,
        name: test.text,
        articleNumber: 'ART-001',
        unitType: 'EA',
        unitQuantity: 1,
        unitNetPrice: 100,
        vatPercentage: 19
      }];
      
      try {
        // Test both UBL and CII formats
        for (const format of ['ubl', 'cii'] as const) {
          const xmlString = await einvoice.toXmlString(format);
          
          // Check if text is preserved
          const preserved = xmlString.includes(test.text);
          console.log(`Encoding test ${test.encoding} in ${format}: ${preserved ? 'preserved' : 'modified'}`);
          
          // Import back
          const newInvoice = new EInvoice();
          await newInvoice.fromXmlString(xmlString);
          
          const descPreserved = newInvoice.subject === test.text;
          console.log(`Encoding test ${test.encoding} round-trip in ${format}: ${descPreserved ? 'success' : 'failed'}`);
        }
      } catch (error) {
        console.log(`Encoding test ${test.encoding} failed: ${error.message}`);
      }
    }
  });

  // Test 3: Emoji and pictographic characters
  await PerformanceTracker.track('emoji-and-pictographs', async () => {
    const emojiTests = [
      {
        name: 'basic-emoji',
        content: 'Invoice 📧 sent ✅'
      },
      {
        name: 'flag-emoji',
        content: 'Country: 🇺🇸 🇬🇧 🇩🇪 🇫🇷'
      },
      {
        name: 'skin-tone-emoji',
        content: 'Approved by 👍🏻👍🏼👍🏽👍🏾👍🏿'
      },
      {
        name: 'zwj-sequences',
        content: 'Family: 👨‍👩‍👧‍👦'
      },
      {
        name: 'mixed-emoji-text',
        content: '💰 Total: €1,234.56 💶'
      }
    ];
    
    for (const test of emojiTests) {
      const einvoice = new EInvoice();
      einvoice.issueDate = new Date(2024, 0, 1);
      einvoice.invoiceId = 'EMOJI-001';
      einvoice.subject = test.content;
      
      einvoice.from = {
        type: 'company',
        name: 'Emoji Company',
        description: test.content,
        address: {
          streetName: 'Emoji Street',
          houseNumber: '1',
          postalCode: '12345',
          city: 'Emoji 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: 'Emoji',
        surname: 'Customer',
        salutation: 'Mr' as const,
        sex: 'male' as const,
        title: 'Doctor' as const,
        description: 'Customer who likes emojis',
        address: {
          streetName: 'Customer Street',
          houseNumber: '2',
          postalCode: '54321',
          city: 'Customer City',
          country: 'DE'
        }
      };
      
      einvoice.items = [{
        position: 1,
        name: test.content,
        articleNumber: 'EMOJI-001',
        unitType: 'EA',
        unitQuantity: 1,
        unitNetPrice: 100,
        vatPercentage: 19
      }];
      
      try {
        const ublString = await einvoice.toXmlString('ubl');
        
        // Check if emoji content is preserved or encoded
        const preserved = ublString.includes(test.content);
        console.log(`Emoji test ${test.name}: ${preserved ? 'preserved' : 'encoded'}`);
        
        // Count grapheme clusters (visual characters)
        const graphemeCount = [...new Intl.Segmenter().segment(test.content)].length;
        console.log(`Emoji test ${test.name} has ${graphemeCount} visual characters`);
      } catch (error) {
        console.log(`Emoji test ${test.name} failed: ${error.message}`);
      }
    }
  });

  // Test 4: Legacy and exotic scripts
  await PerformanceTracker.track('exotic-scripts', async () => {
    const scripts = [
      { name: 'chinese-traditional', text: '發票編號：貳零貳肆' },
      { name: 'japanese-mixed', text: '請求書番号：２０２４年' },
      { name: 'korean', text: '송장 번호: 2024' },
      { name: 'thai', text: 'ใบแจ้งหนี้: ๒๐๒๔' },
      { name: 'devanagari', text: 'चालान संख्या: २०२४' },
      { name: 'bengali', text: 'চালান নং: ২০২৪' },
      { name: 'tamil', text: 'விலைப்பட்டியல்: ௨௦௨௪' }
    ];
    
    for (const script of scripts) {
      const einvoice = new EInvoice();
      einvoice.issueDate = new Date(2024, 0, 1);
      einvoice.invoiceId = `SCRIPT-${script.name}`;
      einvoice.subject = script.text;
      
      einvoice.from = {
        type: 'company',
        name: 'International Company',
        description: script.text,
        address: {
          streetName: 'International Street',
          houseNumber: '1',
          postalCode: '12345',
          city: 'International City',
          country: 'DE'
        },
        status: 'active',
        foundedDate: { year: 2020, month: 1, day: 1 },
        registrationDetails: {
          vatId: 'DE123456789',
          registrationId: 'HRB 12345',
          registrationName: 'Commercial Register'
        }
      };
      
      einvoice.to = {
        type: 'company',
        name: 'Local Company',
        description: 'Customer',
        address: {
          streetName: 'Local Street',
          houseNumber: '2',
          postalCode: '54321',
          city: 'Local City',
          country: 'DE'
        },
        status: 'active',
        foundedDate: { year: 2019, month: 1, day: 1 },
        registrationDetails: {
          vatId: 'DE987654321',
          registrationId: 'HRB 54321',
          registrationName: 'Commercial Register'
        }
      };
      
      einvoice.items = [{
        position: 1,
        name: script.text,
        articleNumber: 'INT-001',
        unitType: 'EA',
        unitQuantity: 1,
        unitNetPrice: 100,
        vatPercentage: 19
      }];
      
      try {
        const ciiString = await einvoice.toXmlString('cii');
        
        const preserved = ciiString.includes(script.text);
        console.log(`Script ${script.name}: ${preserved ? 'preserved' : 'encoded'}`);
        
        // Test round-trip
        const newInvoice = new EInvoice();
        await newInvoice.fromXmlString(ciiString);
        
        const descPreserved = newInvoice.subject === script.text;
        console.log(`Script ${script.name} round-trip: ${descPreserved ? 'success' : 'modified'}`);
      } catch (error) {
        console.log(`Script ${script.name} failed: ${error.message}`);
      }
    }
  });

  // Test 5: XML special characters in unusual positions
  await PerformanceTracker.track('xml-special-characters', async () => {
    const specialChars = [
      { char: '<', desc: 'less than' },
      { char: '>', desc: 'greater than' },
      { char: '&', desc: 'ampersand' },
      { char: '"', desc: 'quote' },
      { char: "'", desc: 'apostrophe' }
    ];
    
    for (const special of specialChars) {
      const einvoice = new EInvoice();
      einvoice.issueDate = new Date(2024, 0, 1);
      einvoice.invoiceId = `XML-${special.desc}`;
      einvoice.subject = `Price ${special.char} 100`;
      
      einvoice.from = {
        type: 'company',
        name: `Company ${special.char} Test`,
        description: 'Special char test',
        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: `Item ${special.char} Test`,
        articleNumber: 'SPEC-001',
        unitType: 'EA',
        unitQuantity: 1,
        unitNetPrice: 100,
        vatPercentage: 19
      }];
      
      try {
        const xmlString = await einvoice.toXmlString('ubl');
        
        // Check if special chars are properly escaped
        const escaped = xmlString.includes(`&${special.desc.replace(' ', '')};`) || 
                       xmlString.includes(`&#${special.char.charCodeAt(0)};`);
        console.log(`XML special ${special.desc}: ${escaped ? 'properly escaped' : 'check encoding'}`);
      } catch (error) {
        console.log(`XML special ${special.desc} failed: ${error.message}`);
      }
    }
  });

  // Test 6: Character set conversion in format transformation
  await PerformanceTracker.track('format-transform-charsets', async () => {
    const testContents = [
      { name: 'multilingual', text: 'Hello مرحبا 你好 Здравствуйте' },
      { name: 'symbols', text: '€ £ ¥ $ ₹ ₽ ¢ ₩' },
      { name: 'accented', text: 'àáäâ èéëê ìíïî òóöô ùúüû ñç' },
      { name: 'mixed-emoji', text: 'Invoice 📄 Total: 💰 Status: ✅' }
    ];
    
    for (const content of testContents) {
      const einvoice = new EInvoice();
      einvoice.issueDate = new Date(2024, 0, 1);
      einvoice.invoiceId = 'CHARSET-001';
      einvoice.subject = content.text;
      
      einvoice.from = {
        type: 'company',
        name: 'Charset Test Company',
        description: content.text,
        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: 'company',
        name: 'Customer Company',
        description: 'Customer',
        address: {
          streetName: 'Customer Street',
          houseNumber: '2',
          postalCode: '54321',
          city: 'Customer City',
          country: 'DE'
        },
        status: 'active',
        foundedDate: { year: 2019, month: 1, day: 1 },
        registrationDetails: {
          vatId: 'DE987654321',
          registrationId: 'HRB 54321',
          registrationName: 'Commercial Register'
        }
      };
      
      einvoice.items = [{
        position: 1,
        name: content.text,
        articleNumber: 'CHARSET-001',
        unitType: 'EA',
        unitQuantity: 1,
        unitNetPrice: 100,
        vatPercentage: 19
      }];
      
      try {
        // Convert from UBL to CII
        const ublString = await einvoice.toXmlString('ubl');
        const newInvoice = new EInvoice();
        await newInvoice.fromXmlString(ublString);
        const ciiString = await newInvoice.toXmlString('cii');
        
        // Check if content was preserved through transformation
        const preserved = ciiString.includes(content.text);
        console.log(`Format transform ${content.name}: ${preserved ? 'preserved' : 'modified'}`);
        
        // Double check with round trip
        const finalInvoice = new EInvoice();
        await finalInvoice.fromXmlString(ciiString);
        const roundTripPreserved = finalInvoice.subject === content.text;
        console.log(`Format transform ${content.name} round-trip: ${roundTripPreserved ? 'success' : 'failed'}`);
      } catch (error) {
        console.log(`Format transform ${content.name} failed: ${error.message}`);
      }
    }
  });
});

// Run the test
tap.start();