murasaki-core/test/sources/kanjidic.test.ts

import { describe, it, afterAll } from 'vitest';
import { createReadStream, createWriteStream } from 'fs';
import { unlink } from 'fs/promises';
import { createDownloader, createXmlToJsonLines, Kanjidic } from '../../src';

const DOWNLOAD_FILENAME = 'kanjidic2.xml' as const;
const PROCESS_OUTPUT_FILENAME = 'kanjidic2.jsonl' as const;

// FIXME use tests that do not download the source data

describe.skip('kanjidic', () => {
  afterAll(async () => {
    try {
      await unlink(DOWNLOAD_FILENAME);
    } catch {
      // noop
    }
  });

  afterAll(async () => {
    try {
      await unlink(PROCESS_OUTPUT_FILENAME);
    } catch {
      // noop
    }
  });

  describe('downloader', () => {
    it('downloads the source data', async () => {
      const readStream = await createDownloader({
        type: Kanjidic.SOURCE_ID,
      });

      return new Promise<void>((resolve, reject) => {
        const out = createWriteStream(DOWNLOAD_FILENAME);

        readStream.on('error', (err) => {
          reject(err);
        });

        out.on('error', (err) => {
          reject(err);
        });

        out.on('finish', () => {
          resolve();
        });

        readStream.pipe(out);
      });
    });
  });

  describe('converter', () => {
    it('converts XML to JSON', () => new Promise<void>((resolve, reject) => {
      const readStream = createReadStream(DOWNLOAD_FILENAME, { encoding: 'utf-8' });

      readStream.on('error', (err) => {
        reject(err);
      });

      const transform = createXmlToJsonLines({
        entryTagName: 'character',
      });

      transform.on('error', (err) => {
        reject(err);
      });

      const out = createWriteStream(PROCESS_OUTPUT_FILENAME);

      out.on('error', (err) => {
        reject(err);
      });

      out.on('finish', () => {
        resolve();
      });

      readStream
        .pipe(transform)
        .pipe(out);
    }));
  });
});