test/test.js

const { execSync } = require('child_process');
const exp = require('constants');

describe('conllu2cmc', () => {
  test('Test sparse mode', (done) => {
    // Modify the command based on your script's location and options
    const command = 'node src/index.js -s < test/data/ndy.conllu';
    const stdout = execSync(command).toString();
    expect(stdout).toContain('😂\t_\t_\tEMOIMG');
    var emoimg_count = (stdout.match(/EMOIMG/g) || []).length;
    expect(emoimg_count).toBe(191);
    var ascimg_count = (stdout.match(/EMOASC/g) || []).length;
    expect(ascimg_count).toBe(30);
    var url_count = (stdout.match(/\tURL\t/g) || []).length;
    expect(url_count).toBe(4);
    var adr_count = (stdout.match(/\tADR\t/g) || []).length;
    expect(adr_count).toBe(1);
    var eot_count = (stdout.match(/\n# eot/g) || []).length;
    expect(eot_count).toBe(1);
    var eof_count = (stdout.match(/\n# eof/g) || []).length;
    expect(eof_count).toBe(1);
    var lines_count = (stdout.split("\n")).length;
    expect(lines_count).toBe(746);
    done();
  });

  test('Test full mode', (done) => {
    const command = 'node src/index.js < test/data/ndy.conllu';
    const stdout = execSync(command).toString();
      expect(stdout).toContain('😂\t_\t_\tEMOIMG');
      var emoimg_count = (stdout.match(/EMOIMG/g) || []).length;
      expect(emoimg_count).toBe(191);
      var ascimg_count = (stdout.match(/EMOASC/g) || []).length;
      expect(ascimg_count).toBe(30);
      var url_count = (stdout.match(/\tURL\t/g) || []).length;
      expect(url_count).toBe(4);
      var adr_count = (stdout.match(/\tADR\t/g) || []).length;
      expect(adr_count).toBe(1);
      var lines_count = (stdout.split("\n")).length;
      expect(lines_count).toBe(6202);
      done();
  });

  test('Regression test for issue #113: emoji modifiers and ZWJ', (done) => {
    // Test that compound emojis with modifiers and ZWJ are recognized as single EMOIMG tokens
    const testInput = `# foundry = base
# text_id = test-113
# text = ✊🏿 and 👨‍👨‍👦
1	✊🏿	_	_	_	_	_	_	_	_
2	and	_	CCONJ	_	_	_	_	_	_
3	👨‍👨‍👦	_	_	_	_	_	_	_	_

`;
    const { execSync } = require('child_process');
    const stdout = execSync('node src/index.js', { input: testInput }).toString();
    
    // Check that compound emojis are tagged as EMOIMG in XPOS column only
    expect(stdout).toContain('✊🏿\t_\t_\tEMOIMG');
    expect(stdout).toContain('👨‍👨‍👦\t_\t_\tEMOIMG');
    
    // Count EMOIMG occurrences (should be 1 per emoji - only in XPOS column)
    var emoimg_count = (stdout.match(/EMOIMG/g) || []).length;
    expect(emoimg_count).toBe(2); // 2 emojis × 1 column = 2
    done();
  });

  test('Regression test for issue #114: Wikipedia emoji templates', (done) => {
    // Test that Wikipedia emoji templates are recognized as EMOWIKI tokens
    const testInput = `# foundry = base
# text_id = test-114
# text = [_EMOJI:{{S|;)}}_] and [_EMOJI:{{cool}}_]
1	[_EMOJI:{{S|;)}}_]	_	_	_	_	_	_	_	_
2	and	_	CCONJ	_	_	_	_	_	_
3	[_EMOJI:{{cool}}_]	_	_	_	_	_	_	_	_

`;
    const { execSync } = require('child_process');
    const stdout = execSync('node src/index.js', { input: testInput }).toString();
    
    // Check that Wikipedia emoji templates are tagged as EMOWIKI in XPOS column only
    expect(stdout).toContain('[_EMOJI:{{S|;)}}_]\t_\t_\tEMOWIKI');
    expect(stdout).toContain('[_EMOJI:{{cool}}_]\t_\t_\tEMOWIKI');
    
    // Count EMOWIKI occurrences (should be 1 per template - only in XPOS column)
    var emowiki_count = (stdout.match(/EMOWIKI/g) || []).length;
    expect(emowiki_count).toBe(2); // 2 templates × 1 column = 2
    done();
  });
});