Files
pn-new-crm/tests/unit/ocr/parse-receipt-text.test.ts

102 lines
2.5 KiB
TypeScript
Raw Normal View History

import { describe, it, expect } from 'vitest';
import { parseReceiptText } from '@/lib/ocr/parse-receipt-text';
describe('parseReceiptText', () => {
it('extracts vendor, date, total, currency, and line items from a basic English receipt', () => {
const text = `
Marina Fuel Station
123 Harbor Way
Anguilla
Date: 2026-04-28
Diesel 45.00
Pump Fee 5.00
Subtotal 50.00
Tax 5.00
TOTAL 55.00
Thank you!
`;
const r = parseReceiptText({ text, ocrConfidence: 92 });
expect(r.establishment).toBe('Marina Fuel Station');
expect(r.date).toBe('2026-04-28');
expect(r.amount).toBe(55);
expect(r.lineItems.length).toBeGreaterThanOrEqual(1);
expect(r.lineItems[0]!.description).toMatch(/diesel/i);
expect(r.confidence).toBeGreaterThan(0.5);
});
it('parses European date and comma-decimal amount', () => {
const text = `
Boulangerie du Port
Rue de la Marina
28/04/2026
Pain 3,50
Café 2,50
Total: 6,00
`;
const r = parseReceiptText({ text });
expect(r.date).toBe('2026-04-28');
expect(r.amount).toBe(6);
expect(r.currency).toBe('EUR');
});
it('handles ISO currency codes when no symbol is present', () => {
const text = `
Receipt
Acme Co.
Total 199.00 USD
`;
const r = parseReceiptText({ text });
expect(r.currency).toBe('USD');
});
it('returns null fields and zero confidence when nothing parseable', () => {
const r = parseReceiptText({ text: '@@@\n!!!\n###' });
expect(r.establishment).toBeNull();
expect(r.amount).toBeNull();
expect(r.date).toBeNull();
expect(r.confidence).toBe(0);
});
it('prefers grand total over subtotal even when subtotal is named "total"', () => {
const text = `
Vendor X
Item A 10.00
Item B 20.00
Subtotal 30.00
Tax 3.00
Grand Total 33.00
`;
const r = parseReceiptText({ text });
expect(r.amount).toBe(33);
});
it('skips obvious total/subtotal lines when extracting line items', () => {
const text = `
Boutique
Shirt 25.00
Tie 15.00
Subtotal 40.00
Tax 4.00
Total 44.00
`;
const r = parseReceiptText({ text });
const descriptions = r.lineItems.map((li) => li.description.toLowerCase());
expect(descriptions.some((d) => d.includes('subtotal'))).toBe(false);
expect(descriptions.some((d) => d.includes('total'))).toBe(false);
expect(descriptions.some((d) => d.includes('tax'))).toBe(false);
});
it('rejects implausible dates', () => {
const r = parseReceiptText({ text: 'Random 1899-04-12 noise' });
expect(r.date).toBeNull();
});
});