import { expect, tap } from '@git.zone/tstest/tapbundle';
import * as plugins from '../plugins.js';
import { EInvoice } from '../../../ts/index.js';
import { CorpusLoader } from '../corpus.loader.js';
import { PerformanceTracker } from '../performance.tracker.js';
tap.test('ENC-05: Special Characters - should handle special and international characters correctly', async (t) => {
// ENC-05: Verify handling of special characters across different languages and scripts
// This test ensures proper support for international invoicing
const performanceTracker = new PerformanceTracker('ENC-05: Special Characters');
const corpusLoader = new CorpusLoader();
t.test('European special characters', async () => {
const startTime = performance.now();
const xmlContent = `
2.1
EU-SPECIAL-CHARS
2025-01-25
European chars test
Åsa Öberg AB (Sweden)
Østergade 42
København
DK
Müller & Schäfer GmbH
Hauptstraße 15
Düsseldorf
DE
François Lefèvre
f.lefevre@müller-schäfer.de
-
Château Margaux (Bordeaux)
Vin rouge, millésime 2015, cépage: Cabernet Sauvignon
-
Prošek (Croatian dessert wine)
Vino desertno, područje: Dalmacija
-
Żubrówka (Polish vodka)
Wódka żytnia z trawą żubrową
`;
const einvoice = new EInvoice();
await einvoice.loadFromString(xmlContent);
const xmlString = einvoice.getXmlString();
// Nordic characters
expect(xmlString).toContain('Åsa Öberg');
expect(xmlString).toContain('Østergade');
expect(xmlString).toContain('København');
// German characters
expect(xmlString).toContain('Müller & Schäfer');
expect(xmlString).toContain('Hauptstraße');
expect(xmlString).toContain('Düsseldorf');
expect(xmlString).toContain('müller-schäfer.de');
// French characters
expect(xmlString).toContain('François Lefèvre');
expect(xmlString).toContain('Château Margaux');
expect(xmlString).toContain('millésime');
expect(xmlString).toContain('cépage');
// Croatian characters
expect(xmlString).toContain('Prošek');
expect(xmlString).toContain('područje');
// Polish characters
expect(xmlString).toContain('Żubrówka');
expect(xmlString).toContain('żytnia');
expect(xmlString).toContain('żubrową');
const elapsed = performance.now() - startTime;
performanceTracker.addMeasurement('european-chars', elapsed);
});
t.test('Currency and monetary symbols', async () => {
const startTime = performance.now();
const xmlContent = `
2.1
CURRENCY-SYMBOLS
Currency symbols: € £ $ ¥ ₹ ₽ ₪ ₩ ₡ ₦ ₨ ₱ ₴ ₵ ₸ ₹ ₺ ₼
€1,234.56
£987.65
$2,345.67
¥123,456
₹98,765
false
Discount (5% off orders > €500)
25.50
Accepted: € EUR, £ GBP, $ USD, ¥ JPY, ₹ INR
`;
const einvoice = new EInvoice();
await einvoice.loadFromString(xmlContent);
const xmlString = einvoice.getXmlString();
// Major currency symbols
expect(xmlString).toContain('€'); // Euro
expect(xmlString).toContain('£'); // Pound
expect(xmlString).toContain('$'); // Dollar
expect(xmlString).toContain('¥'); // Yen
expect(xmlString).toContain('₹'); // Rupee
expect(xmlString).toContain('₽'); // Ruble
expect(xmlString).toContain('₪'); // Shekel
expect(xmlString).toContain('₩'); // Won
// Verify monetary formatting
expect(xmlString).toContain('€1,234.56');
expect(xmlString).toContain('£987.65');
expect(xmlString).toContain('$2,345.67');
expect(xmlString).toContain('¥123,456');
expect(xmlString).toContain('₹98,765');
const elapsed = performance.now() - startTime;
performanceTracker.addMeasurement('currency-symbols', elapsed);
});
t.test('Mathematical and technical symbols', async () => {
const startTime = performance.now();
const xmlContent = `
2.1
MATH-SYMBOLS
Math symbols: ± × ÷ ≤ ≥ ≠ ≈ ∞ √ ∑ ∏ ∫ ∂ ∇ ∈ ∉ ⊂ ⊃ ∪ ∩
100.00
95.00
Discount ≥ 10 units
-
Precision tool ± 0.001mm
Temperature range
-40°C ≤ T ≤ +85°C
Dimensions
10cm × 5cm × 2cm
-
√2 ≈ 1.414, π ≈ 3.14159, e ≈ 2.71828
Formula
Area = πr² (where r = radius)
`;
const einvoice = new EInvoice();
await einvoice.loadFromString(xmlContent);
const xmlString = einvoice.getXmlString();
// Mathematical operators
expect(xmlString).toContain('±'); // Plus-minus
expect(xmlString).toContain('×'); // Multiplication
expect(xmlString).toContain('÷'); // Division
expect(xmlString).toContain('≤'); // Less than or equal
expect(xmlString).toContain('≥'); // Greater than or equal
expect(xmlString).toContain('≠'); // Not equal
expect(xmlString).toContain('≈'); // Approximately
expect(xmlString).toContain('∞'); // Infinity
expect(xmlString).toContain('√'); // Square root
expect(xmlString).toContain('π'); // Pi
expect(xmlString).toContain('°'); // Degree
const elapsed = performance.now() - startTime;
performanceTracker.addMeasurement('math-symbols', elapsed);
});
t.test('Asian scripts and characters', async () => {
const startTime = performance.now();
const xmlContent = `
2.1
ASIAN-SCRIPTS
株式会社山田商事 (Yamada Trading Co., Ltd.)
東京都千代田区丸の内1-1-1
東京
JP
北京科技有限公司 (Beijing Tech Co., Ltd.)
北京市朝阳区建国路88号
北京
CN
-
전자제품 (Electronics)
최신 스마트폰 모델
-
कंप्यूटर उपकरण
नवीनतम लैपटॉप मॉडल
-
ซอฟต์แวร์คอมพิวเตอร์
โปรแกรมสำนักงาน
`;
const einvoice = new EInvoice();
await einvoice.loadFromString(xmlContent);
const xmlString = einvoice.getXmlString();
// Japanese (Kanji, Hiragana, Katakana)
expect(xmlString).toContain('株式会社山田商事');
expect(xmlString).toContain('東京都千代田区丸の内');
// Chinese (Simplified)
expect(xmlString).toContain('北京科技有限公司');
expect(xmlString).toContain('北京市朝阳区建国路');
// Korean (Hangul)
expect(xmlString).toContain('전자제품');
expect(xmlString).toContain('최신 스마트폰 모델');
// Hindi (Devanagari)
expect(xmlString).toContain('कंप्यूटर उपकरण');
expect(xmlString).toContain('नवीनतम लैपटॉप मॉडल');
// Thai
expect(xmlString).toContain('ซอฟต์แวร์คอมพิวเตอร์');
expect(xmlString).toContain('โปรแกรมสำนักงาน');
const elapsed = performance.now() - startTime;
performanceTracker.addMeasurement('asian-scripts', elapsed);
});
t.test('Arabic and RTL scripts', async () => {
const startTime = performance.now();
const xmlContent = `
2.1
RTL-SCRIPTS
شركة التقنية المحدودة
شارع الملك فهد
الرياض
SA
חברת הטכנולוגיה בע"מ
רחוב דיזנגוף 123
תל אביב
IL
الدفع: 30 يومًا صافي
-
منتج إلكتروني
جهاز كمبيوتر محمول
-
מוצר אלקטרוני
מחשב נייד מתקדם
`;
const einvoice = new EInvoice();
await einvoice.loadFromString(xmlContent);
const xmlString = einvoice.getXmlString();
// Arabic
expect(xmlString).toContain('شركة التقنية المحدودة');
expect(xmlString).toContain('شارع الملك فهد');
expect(xmlString).toContain('الرياض');
expect(xmlString).toContain('الدفع: 30 يومًا صافي');
expect(xmlString).toContain('منتج إلكتروني');
// Hebrew
expect(xmlString).toContain('חברת הטכנולוגיה בע"מ');
expect(xmlString).toContain('רחוב דיזנגוף');
expect(xmlString).toContain('תל אביב');
expect(xmlString).toContain('מוצר אלקטרוני');
const elapsed = performance.now() - startTime;
performanceTracker.addMeasurement('rtl-scripts', elapsed);
});
t.test('Emoji and emoticons', async () => {
const startTime = performance.now();
const xmlContent = `
2.1
EMOJI-TEST
Thank you for your order! 😊 🎉 🚀
Payment methods: 💳 💰 🏦
-
Premium Package 🌟
Includes: 📱 💻 🖱️ ⌨️ 🎧
-
Express Shipping 🚚💨
Delivery: 📦 → 🏠 (1-2 days)
-
Customer Support 24/7 ☎️
Contact: 📧 📞 💬
`;
const einvoice = new EInvoice();
await einvoice.loadFromString(xmlContent);
const xmlString = einvoice.getXmlString();
// Common emojis
expect(xmlString).toContain('😊'); // Smiling face
expect(xmlString).toContain('🎉'); // Party
expect(xmlString).toContain('🚀'); // Rocket
expect(xmlString).toContain('💳'); // Credit card
expect(xmlString).toContain('💰'); // Money bag
expect(xmlString).toContain('🏦'); // Bank
expect(xmlString).toContain('🌟'); // Star
expect(xmlString).toContain('📱'); // Phone
expect(xmlString).toContain('💻'); // Laptop
expect(xmlString).toContain('🚚'); // Truck
expect(xmlString).toContain('📦'); // Package
expect(xmlString).toContain('🏠'); // House
expect(xmlString).toContain('☎️'); // Phone
expect(xmlString).toContain('📧'); // Email
expect(xmlString).toContain('💬'); // Chat
const elapsed = performance.now() - startTime;
performanceTracker.addMeasurement('emoji', elapsed);
});
t.test('Corpus special character validation', async () => {
const startTime = performance.now();
let processedCount = 0;
let specialCharCount = 0;
const specialCharFiles: string[] = [];
const files = await corpusLoader.getAllFiles();
const xmlFiles = files.filter(f => f.endsWith('.xml'));
// Check sample for special characters
const sampleSize = Math.min(60, xmlFiles.length);
const sample = xmlFiles.slice(0, sampleSize);
for (const file of sample) {
try {
const content = await corpusLoader.readFile(file);
const einvoice = new EInvoice();
if (typeof content === 'string') {
await einvoice.loadFromString(content);
} else {
await einvoice.loadFromBuffer(content);
}
const xmlString = einvoice.getXmlString();
// Check for non-ASCII characters
if (/[^\x00-\x7F]/.test(xmlString)) {
specialCharCount++;
// Check for specific character ranges
if (/[À-ÿ]/.test(xmlString)) {
specialCharFiles.push(`${file} (Latin Extended)`);
} else if (/[Ā-ſ]/.test(xmlString)) {
specialCharFiles.push(`${file} (Latin Extended-A)`);
} else if (/[\u0400-\u04FF]/.test(xmlString)) {
specialCharFiles.push(`${file} (Cyrillic)`);
} else if (/[\u4E00-\u9FFF]/.test(xmlString)) {
specialCharFiles.push(`${file} (CJK)`);
} else if (/[\u0600-\u06FF]/.test(xmlString)) {
specialCharFiles.push(`${file} (Arabic)`);
}
}
processedCount++;
} catch (error) {
console.log(`Special char issue in ${file}:`, error.message);
}
}
console.log(`Special character corpus test: ${specialCharCount}/${processedCount} files contain special characters`);
if (specialCharFiles.length > 0) {
console.log('Sample files with special characters:', specialCharFiles.slice(0, 5));
}
expect(processedCount).toBeGreaterThan(0);
const elapsed = performance.now() - startTime;
performanceTracker.addMeasurement('corpus-special', elapsed);
});
t.test('Zero-width and invisible characters', async () => {
const startTime = performance.now();
const xmlContent = `
2.1
INVISIBLE-CHARS
Zero-widthspace (U+200B)
Nonbreakingzerowidthjoiner
-
Softhyphentest
Left‐to‐rightmark and right‐to‐leftmark
`;
const einvoice = new EInvoice();
await einvoice.loadFromString(xmlContent);
const xmlString = einvoice.getXmlString();
// These characters might be preserved or stripped
// Check that the text is still readable
expect(xmlString).toMatch(/Zero.*width.*space/);
expect(xmlString).toMatch(/Non.*breaking.*zero.*width.*joiner/);
expect(xmlString).toMatch(/Soft.*hyphen.*test/);
const elapsed = performance.now() - startTime;
performanceTracker.addMeasurement('invisible-chars', elapsed);
});
// Print performance summary
performanceTracker.printSummary();
// Performance assertions
const avgTime = performanceTracker.getAverageTime();
expect(avgTime).toBeLessThan(150); // Special character operations should be reasonably fast
});
tap.start();