jadb/lib/_data_ingestion/kanjidic/xml_parser.dart

import 'package:jadb/_data_ingestion/kanjidic/objects.dart';
import 'package:jadb/util/romaji_transliteration.dart';
import 'package:xml/xml.dart';

List<Character> parseKANJIDICData(XmlElement root) {
  final List<Character> result = [];
  for (final c in root.findElements('character')) {
    final kanji = c.findElements('literal').first.innerText;

    final codepoint = c.findElements('codepoint').firstOrNull;
    final radical = c.findElements('radical').firstOrNull;
    final misc = c.findElements('misc').first;
    final dicNumber = c.findElements('dic_number').firstOrNull;
    final queryCode = c.findElements('query_code').first;
    final readingMeaning = c.findElements('reading_meaning').firstOrNull;

    // TODO: Group readings and meanings by their rmgroup parent node.

    result.add(
      Character(
        literal: kanji,
        strokeCount: int.parse(
          misc.findElements('stroke_count').first.innerText,
        ),
        grade: int.tryParse(
          misc.findElements('grade').firstOrNull?.innerText ?? '',
        ),
        frequency: int.tryParse(
          misc.findElements('freq').firstOrNull?.innerText ?? '',
        ),
        jlpt: int.tryParse(
          misc.findElements('jlpt').firstOrNull?.innerText ?? '',
        ),
        radicalName: misc
            .findElements('rad_name')
            .map((e) => e.innerText)
            .toList(),
        codepoints:
            codepoint
                ?.findElements('cp_value')
                .map(
                  (e) => CodePoint(
                    kanji: kanji,
                    type: e.getAttribute('cp_type')!,
                    codepoint: e.innerText,
                  ),
                )
                .toList() ??
            [],
        radical: radical
            ?.findElements('rad_value')
            .where((e) => e.getAttribute('rad_type') == 'classical')
            .map(
              (e) => Radical(kanji: kanji, radicalId: int.parse(e.innerText)),
            )
            .firstOrNull,
        strokeMiscounts: misc
            .findElements('stroke_count')
            .skip(1)
            .map((e) => int.parse(e.innerText))
            .toList(),
        variants: misc
            .findElements('variant')
            .map(
              (e) => Variant(
                kanji: kanji,
                type: e.getAttribute('var_type')!,
                variant: e.innerText,
              ),
            )
            .toList(),
        dictionaryReferences:
            dicNumber
                ?.findElements('dic_ref')
                .where((e) => e.getAttribute('dr_type') != 'moro')
                .map(
                  (e) => DictionaryReference(
                    kanji: kanji,
                    type: e.getAttribute('dr_type')!,
                    ref: e.innerText,
                  ),
                )
                .toList() ??
            [],
        dictionaryReferencesMoro:
            dicNumber
                ?.findElements('dic_ref')
                .where((e) => e.getAttribute('dr_type') == 'moro')
                .map(
                  (e) => DictionaryReferenceMoro(
                    kanji: kanji,
                    ref: e.innerText,
                    page: int.tryParse(e.getAttribute('m_page') ?? ''),
                    volume: int.tryParse(e.getAttribute('m_vol') ?? ''),
                  ),
                )
                .toList() ??
            [],
        querycodes: queryCode
            .findElements('q_code')
            .map(
              (e) => QueryCode(
                kanji: kanji,
                code: e.innerText,
                type: e.getAttribute('qc_type')!,
                skipMisclassification: e.getAttribute('skip_misclass'),
              ),
            )
            .toList(),
        readings:
            readingMeaning
                ?.findAllElements('reading')
                .where(
                  (e) =>
                      !['ja_on', 'ja_kun'].contains(e.getAttribute('r_type')),
                )
                .map(
                  (e) => Reading(
                    kanji: kanji,
                    type: e.getAttribute('r_type')!,
                    reading: e.innerText,
                  ),
                )
                .toList() ??
            [],
        kunyomi:
            readingMeaning
                ?.findAllElements('reading')
                .where((e) => e.getAttribute('r_type') == 'ja_kun')
                .map(
                  (e) => Kunyomi(
                    kanji: kanji,
                    yomi: e.innerText,
                    isJouyou: e.getAttribute('r_status') == 'jy',
                  ),
                )
                .toList() ??
            [],
        onyomi:
            readingMeaning
                ?.findAllElements('reading')
                .where((e) => e.getAttribute('r_type') == 'ja_on')
                .map(
                  (e) => Onyomi(
                    kanji: kanji,
                    yomi: transliterateKatakanaToHiragana(e.innerText),
                    isJouyou: e.getAttribute('r_status') == 'jy',
                    type: e.getAttribute('on_type'),
                  ),
                )
                .toList() ??
            [],
        meanings:
            readingMeaning
                ?.findAllElements('meaning')
                .map(
                  (e) => Meaning(
                    kanji: kanji,
                    language: e.getAttribute('m_lang') ?? 'eng',
                    meaning: e.innerText,
                  ),
                )
                .toList() ??
            [],
        nanori:
            readingMeaning
                ?.findElements('nanori')
                .map((e) => e.innerText)
                .toList() ??
            [],
      ),
    );
  }
  return result;
}