pn-new-crm/tests/unit/services/document-field-detector.test.ts

import { describe, it, expect, vi } from 'vitest';

// Mock pdfjs-dist before importing the service. The detector calls
// `import('pdfjs-dist/legacy/build/pdf.mjs')` dynamically; we stub the
// module with a fake document whose pages return canned text items so
// we can assert the anchor-matching + coordinate-conversion logic
// without needing a real PDF.
vi.mock('pdfjs-dist/legacy/build/pdf.mjs', () => ({
  getDocument: (_opts: unknown) => ({
    promise: Promise.resolve({
      numPages: 1,
      getPage: async (_n: number) => ({
        getViewport: ({ scale: _s }: { scale: number }) => ({
          width: 595, // A4 in pt
          height: 842,
        }),
        getTextContent: async () => ({
          items: [
            // Item 0: a signature anchor near the bottom-left
            {
              str: 'Signature: ',
              transform: [1, 0, 0, 1, 50, 100],
              width: 70,
            },
            // Item 1: a date anchor next to it
            {
              str: 'Date: ',
              transform: [1, 0, 0, 1, 250, 100],
              width: 40,
            },
            // Item 2: recipient label nearby
            {
              str: 'Buyer',
              transform: [1, 0, 0, 1, 50, 130],
              width: 40,
            },
            // Item 3: unrelated body text (should not match)
            {
              str: 'The parties hereby agree…',
              transform: [1, 0, 0, 1, 50, 200],
              width: 200,
            },
          ],
        }),
      }),
    }),
  }),
}));

import { detectFields } from '@/lib/services/document-field-detector';

describe('detectFields', () => {
  it('returns matches for known anchors with the right type + page', async () => {
    const result = await detectFields(Buffer.from('%PDF-1.7'));
    expect(result.length).toBeGreaterThanOrEqual(2);
    const sig = result.find((r) => r.type === 'SIGNATURE');
    const date = result.find((r) => r.type === 'DATE');
    expect(sig).toBeDefined();
    expect(date).toBeDefined();
    expect(sig?.pageNumber).toBe(1);
    expect(date?.pageNumber).toBe(1);
  });

  it('infers recipient label from nearby text', async () => {
    const result = await detectFields(Buffer.from('%PDF-1.7'));
    const sig = result.find((r) => r.type === 'SIGNATURE');
    expect(sig?.inferredRecipientLabel).toBe('Buyer');
  });

  it('returns percent coordinates in [0, 100]', async () => {
    const result = await detectFields(Buffer.from('%PDF-1.7'));
    for (const f of result) {
      expect(f.pageX).toBeGreaterThanOrEqual(0);
      expect(f.pageX).toBeLessThanOrEqual(100);
      expect(f.pageY).toBeGreaterThanOrEqual(0);
      expect(f.pageY).toBeLessThanOrEqual(100);
      expect(f.pageWidth).toBeGreaterThan(0);
      expect(f.pageHeight).toBeGreaterThan(0);
    }
  });

  it('attaches the anchor text + a confidence score', async () => {
    const result = await detectFields(Buffer.from('%PDF-1.7'));
    const sig = result.find((r) => r.type === 'SIGNATURE');
    expect(sig?.anchorText).toMatch(/signature/i);
    expect(sig?.confidence).toBeGreaterThan(0.5);
    expect(sig?.confidence).toBeLessThanOrEqual(1);
  });

  it('does not match body text that lacks a signing-block keyword', async () => {
    const result = await detectFields(Buffer.from('%PDF-1.7'));
    // The "The parties hereby agree" item should not produce a TEXT
    // detection (no underscore run, no keyword).
    expect(result.find((r) => r.anchorText?.includes('parties'))).toBeUndefined();
  });

  it('gracefully returns [] when pdfjs throws', async () => {
    // Force pdfjs to reject for this one call
    const mod = await import('pdfjs-dist/legacy/build/pdf.mjs');
    const orig = mod.getDocument;
    (mod as unknown as { getDocument: typeof orig }).getDocument = () =>
      ({ promise: Promise.reject(new Error('boom')) }) as ReturnType<typeof orig>;
    const result = await detectFields(Buffer.from('not-a-pdf'));
    expect(result).toEqual([]);
    (mod as unknown as { getDocument: typeof orig }).getDocument = orig;
  });
});