halkyon/halkyon_learn_curator/test/jmdict_xml_test.dart

import 'dart:convert';
import 'dart:io';

import 'package:halkyon_learn_curator/src/sources/jlpt.dart';
import 'package:halkyon_learn_curator/src/sources/jmdict_xml.dart';
import 'package:test/test.dart';
import 'package:xml/xml.dart';

/// [text] as a stream of pieces [size] characters long.
Stream<String> pieces(String text, int size) => Stream.fromIterable([
  for (var at = 0; at < text.length; at += size)
    text.substring(at, (at + size).clamp(0, text.length)),
]);

void main() {
  late String xml;
  late List<Map<String, Object?>> lines;

  setUpAll(() async {
    xml = await File('test/fixtures/jmdict-e-examp-sample.xml').readAsString();
    final jlpt = await readJlptLevels(Directory('test/fixtures/jlpt'));
    lines = [
      for (final line in await jmdictXmlDictionary(
        Stream.value(xml),
        jlpt: jlpt,
        examples: true,
      ).toList())
        jsonDecode(line) as Map<String, Object?>,
    ];
  });

  Map<String, Object?> entry(String id) =>
      lines.skip(1).firstWhere((line) => line['id'] == id);
  List<Map<String, Object?>> senses(String id) => [
    for (final sense in (entry(id)['entry']! as Map)['senses'] as List)
      sense as Map<String, Object?>,
  ];

  test("the header names EDRDG's JMdict and the day it was made, and "
      'credits the sentences and the levels', () {
    final header = lines.first;
    expect(header['source'], 'jmdict');
    expect(header['kind'], 'words');
    expect(header['sourceVersion'], '2026-10-09');
    expect(
      [
        for (final credit in header['credits']! as List)
          (credit as Map)['name'],
      ],
      ['JMdict', 'JLPT vocabulary lists', 'Tatoeba'],
    );
    expect(lines, hasLength(6));
  });

  test('a word as from the JSON: its headword and reading, labels, keys, '
      "and its senses with JMdict's codes as codes", () {
    final taberu = entry('1358280');

    expect((taberu['headword'], taberu['reading']), ('食べる', 'たべる'));
    expect(taberu['labels'], ['common', 'jlpt:n5']);
    expect(taberu['keys'], containsAll(['食べる', 'たべる', 'taberu', 'eat']));
    expect(senses('1358280').first['pos'], ['v1', 'vt']);
  });

  test("a sense's sentences from Tatoeba, the word as it is in each", () {
    expect((senses('1358280').first['examples']! as List).first, {
      'text': '食べる',
      'ja': 'もっと果物を食べるべきです。',
      'en': 'You should eat more fruit.',
      'tatoeba': '193344',
    });
  });

  test('a word usually written in kana has a kana headword; a sense kept to '
      'one reading says so', () {
    final sono = entry('1006830');

    expect(sono['headword'], 'その');
    expect(senses('1006830').first['misc'], ['uk']);
    expect(senses('1006830').last['readings'], ['その']);
  });

  test('the reading is the one restricted to the headword', () {
    expect(entry('1000110')['reading'], 'シーディープレーヤー');
  });

  test(
    'entities cut between two pieces of the stream are read whole',
    () async {
      final again = [
        for (final line in await jmdictXmlDictionary(
          pieces(xml, 7),
          examples: true,
        ).toList())
          jsonDecode(line) as Map<String, Object?>,
      ];
      expect(
        [for (final line in again.skip(1)) line['entry']],
        [for (final line in lines.skip(1)) line['entry']],
      );
    },
  );

  test('a sense without parts of speech has those of the sense before', () {
    final word = jmdictWordOf(
      XmlDocument.parse('''
<entry><ent_seq>1</ent_seq><r_ele><reb>あ</reb></r_ele>
<sense><pos>n</pos><gloss>one</gloss></sense>
<sense><gloss>two</gloss></sense>
<sense><pos>adv</pos><gloss>three</gloss></sense></entry>''').rootElement,
    );
    expect(
      [
        for (final sense in word['sense']! as List)
          (sense as Map)['partOfSpeech'],
      ],
      [
        ['n'],
        ['n'],
        ['adv'],
      ],
    );
  });
}