test/test.js

const { execSync } = require('child_process');
const exp = require('constants');
const packageVersion = require('../package.json').version;

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 hst_count = (stdout.match(/\tHST\t/g) || []).length;
    expect(hst_count).toBe(12);
    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(803);
    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 hst_count = (stdout.match(/\tHST\t/g) || []).length;
    expect(hst_count).toBe(12);
    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('Test version flag', (done) => {
    const stdout = execSync('node src/index.js -V').toString().trim();

    expect(stdout).toBe(packageVersion);
    done();
  });

  test('Test cmc-tagger alias', (done) => {
    const stdout = execSync('npm exec --yes --package=. cmc-tagger -- -V').toString().trim();
    const versionLine = stdout.split('\n').pop().trim();

    expect(versionLine).toBe(packageVersion);
    done();
  });

  test('Regression test for hashtags: emit HST', (done) => {
    const testInput = [
      '# foundry = base',
      '# text_id = test-hashtag',
      '# text = #KorAP #3D #10',
      ['1', '#KorAP', '_', '_', '_', '_', '_', '_', '_', '_'].join('\t'),
      ['2', '#3D', '_', '_', '_', '_', '_', '_', '_', '_'].join('\t'),
      ['3', '#10', '_', '_', '_', '_', '_', '_', '_', '_'].join('\t'),
      ''
    ].join('\n');
    const stdout = execSync('node src/index.js', { input: testInput }).toString();

    expect(stdout).toContain('#KorAP\t_\t_\tHST');
    expect(stdout).toContain('#3D\t_\t_\tHST');
    expect(stdout).not.toContain('#10\t_\t_\tHST');

    var hst_count = (stdout.match(/\tHST\t/g) || []).length;
    expect(hst_count).toBe(2);
    done();
  });

  test('Regression test for hashtags with Unicode letters: emit HST', (done) => {
    const testInput = [
      '# foundry = base',
      '# text_id = test-hashtag-unicode',
      '# text = #okeichhörejetztauf #schön #10',
      ['1', '#okeichhörejetztauf', '_', '_', '_', '_', '_', '_', '_', '_'].join('\t'),
      ['2', '#schön', '_', '_', '_', '_', '_', '_', '_', '_'].join('\t'),
      ['3', '#10', '_', '_', '_', '_', '_', '_', '_', '_'].join('\t'),
      ''
    ].join('\n');
    const stdout = execSync('node src/index.js', { input: testInput }).toString();

    expect(stdout).toContain('#okeichhörejetztauf\t_\t_\tHST');
    expect(stdout).toContain('#schön\t_\t_\tHST');
    expect(stdout).not.toContain('#10\t_\t_\tHST');

    var hst_count = (stdout.match(/\tHST\t/g) || []).length;
    expect(hst_count).toBe(2);
    done();
  });

  test('Regression test for addresses: emit ADR regardless of existing POS values', (done) => {
    const testInput = [
      '# foundry = base',
      '# text_id = test-address',
      '# text = @handle @markup',
      ['1', '@handle', '_', 'PROPN', '_', '_', '_', '_', '_', '_'].join('\t'),
      ['2', '@markup', '_', 'NE', '_', '_', '_', '_', '_', '_'].join('\t'),
      ''
    ].join('\n');
    const stdout = execSync('node src/index.js', { input: testInput }).toString();

    expect(stdout).toContain('@handle\t_\tPROPN\tADR');
    expect(stdout).toContain('@markup\t_\tNE\tADR');

    var adr_count = (stdout.match(/\tADR\t/g) || []).length;
    expect(adr_count).toBe(2);
    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 and lemma has base emoji
    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();
  });

  test('Test emoji metadata in FEATS column', (done) => {
    // Test that EMOIMG tokens have populated FEATS column
    const testInput = `# foundry = base
# text_id = test-feats
# text = 😇
1	😇	_	_	_	_	_	_	_	_

`;
    const { execSync } = require('child_process');
    const stdout = execSync('node src/index.js', { input: testInput }).toString();

    // Check that 😇 has correct metadata
    // g=smileys_and_emotion|s=face_smiling|q=fully_qualified|v=E1.0|n=smiling_face_with_halo
    // Note: spaces in data are replaced by _ in our script
    expect(stdout).toContain('g=smileys_&_emotion|s=face_smiling|q=fully_qualified|v=E1.0|n=smiling_face_with_halo');

    // Also check for the base emoji lemma and tags
    expect(stdout).toContain('😇\t😇\t_\tEMOIMG');

    done();
  });

  test('Test normalized emoji-name separators in FEATS', (done) => {
    const testInput = `# foundry = base
# text_id = test-name-normalization
# text = 👍🏻 👨‍👨‍👦
1	👍🏻	_	_	_	_	_	_	_	_
2	👨‍👨‍👦	_	_	_	_	_	_	_	_

`;
    const stdout = execSync('node src/index.js', { input: testInput }).toString();

    expect(stdout).toContain('n=thumbs_up:light_skin_tone');
    expect(stdout).toContain('n=family:man,man,boy');
    expect(stdout).not.toContain('n=thumbs_up:_light_skin_tone');
    expect(stdout).not.toContain('n=family:_man,_man,_boy');

    done();
  });
});