obdfcascrape/tests/scrape.pl.js

const tape = require('tape');
const _test = require('tape-promise').default; // <---- notice 'default'
const test = _test(tape); // decorate tape

const fs = require('fs');
const jsonfile = require('jsonfile');

const Poland = require('../ncas/pl');

const plScraper = new Poland();

const failure = { 'fail':true };
const empty = {};

test.test('Payment Services', async t => {
  test.test('POLAND:: Process PS Entity 001', async t => {
    const htmlFile = 'tests/data/pl/pi_001.html';
    t.test('🇵🇱::Extract entity details', async t => {
      const psDetail = fs.readFileSync(htmlFile);

      console.log(psDetail);
      const expectedJSON = jsonfile.readFileSync('tests/data/pl/pi_001_detail.json');

      const output = await plScraper.extractEntityDetails(psDetail);

      t.deepEquals(output, expectedJSON, 'Extracted entity details from Page');

      t.end();
    });

    t.test('🇵🇱::Extract entity services', async t => {
      const psDetail = fs.readFileSync(htmlFile);
      const expectedJSON = jsonfile.readFileSync('tests/data/pl/pi_001_services.json');

      const output = await plScraper.extractEntityServices(psDetail);

      t.deepEquals(output, expectedJSON, 'Extracted entity services from Page');

      t.end();
    });

    t.test('🇵🇱::Extract entity Agents', async t => {
      const psDetail = fs.readFileSync(htmlFile);
      const expectedJSON = empty;

      const output = await plScraper.extractEntityAgents(psDetail);

      t.deepEquals(output, expectedJSON, 'Extracted entity agents from Page');
      t.equals(output.length, 0, 'Extracted 0 entity agents from Page');

      t.end();
    });

    t.test('🇵🇱::Extract entity Branches', async t => {
      const psDetail = fs.readFileSync(htmlFile);
      const expectedJSON = empty;

      const output = await plScraper.extractEntityBranches(psDetail);

      t.deepEquals(output, expectedJSON, 'Extracted entity branches from Page');
      t.equals(output.length, 0, 'Extracted 0 entity branches from Page');
      t.end();
    });

    t.test('🇵🇱::Extract entity Activities', async t => {
      const psDetail = fs.readFileSync(htmlFile);
      const expectedJSON = empty;

      const output = await plScraper.extractEntityActivity(psDetail);

      t.deepEquals(output, expectedJSON, 'Extracted entity activities from Page');

      t.end();
    });

    t.end();
  });

  test.test('POLAND:: Process PS Entity 002', async t => {
    const htmlFile = 'tests/data/pl/pi_002.html';

    t.test('🇵🇱::Extract entity details', async t => {
      const psDetail = fs.readFileSync(htmlFile);
      const expectedJSON = jsonfile.readFileSync('tests/data/pl/pi_002_detail.json');

      const output = await plScraper.extractEntityDetails(psDetail);

      t.deepEquals(output, expectedJSON, 'Extracted entity details from Page');

      t.end();
    });

    t.test('🇵🇱::Extract entity services', async t => {
      const psDetail = fs.readFileSync(htmlFile);
      const expectedJSON = jsonfile.readFileSync('tests/data/pl/pi_002_services.json');

      const output = await plScraper.extractEntityServices(psDetail);

      t.deepEquals(output, expectedJSON, 'Extracted entity services from Page');

      t.end();
    });

    t.test('🇵🇱::Extract entity Agents', async t => {
      const psDetail = fs.readFileSync(htmlFile);
      const expectedJSON = jsonfile.readFileSync('tests/data/pl/pi_002_agents.json');

      const output = await plScraper.extractEntityAgents(psDetail);

      t.deepEquals(output, expectedJSON, 'Extracted entity agents from Page');
      t.equals(output.length, 5, 'Extracted 5 entity agents from Page');

      t.end();
    });

    t.test('🇵🇱::Extract entity Branches', async t => {
      const psDetail = fs.readFileSync(htmlFile);
      const expectedJSON = jsonfile.readFileSync('tests/data/pl/pi_002_branches.json');

      const output = await plScraper.extractEntityBranches(psDetail);

      t.deepEquals(output, expectedJSON, 'Extracted entity branches from Page');
      t.equals(output.length, 1, 'Extracted 1 entity branches from Page');
      t.end();
    });

    t.test('🇵🇱::Extract entity Activities', async t => {
      const psDetail = fs.readFileSync(htmlFile);
      const expectedJSON = empty;

      const output = await plScraper.extractEntityActivity(psDetail);

      t.deepEquals(output, expectedJSON, 'Extracted entity activities from Page');

      t.end();
    });

    t.end();
  });

  test.test('POLAND:: Process PS Entity 003', async t => {
    const htmlFile = 'tests/data/pl/pi_003.html';
    t.test('🇵🇱::Extract entity details', async t => {
      const psDetail = fs.readFileSync(htmlFile);
      const expectedJSON = jsonfile.readFileSync('tests/data/pl/pi_003_detail.json');

      const output = await plScraper.extractEntityDetails(psDetail);

      t.deepEquals(output, expectedJSON, 'Extracted entity details from Page');

      t.end();
    });

    t.test('🇵🇱::Extract entity services', async t => {
      const psDetail = fs.readFileSync(htmlFile);
      const expectedJSON = jsonfile.readFileSync('tests/data/pl/pi_003_services.json');

      const output = await plScraper.extractEntityServices(psDetail);

      t.deepEquals(output, expectedJSON, 'Extracted entity services from Page');

      t.end();
    });

    t.test('🇵🇱::Extract entity Agents', async t => {
      const psDetail = fs.readFileSync(htmlFile);
      const expectedJSON = empty;

      const output = await plScraper.extractEntityAgents(psDetail);

      t.deepEquals(output, expectedJSON, 'Extracted entity agents from Page');
      t.equals(output.length, 0, 'Extracted 0 entity agents from Page');

      t.end();
    });

    t.test('🇵🇱::Extract entity Branches', async t => {
      const psDetail = fs.readFileSync(htmlFile);
      const expectedJSON = empty;

      const output = await plScraper.extractEntityBranches(psDetail);

      t.deepEquals(output, expectedJSON, 'Extracted entity branches from Page');
      t.equals(output.length, 0, 'Extracted 0 entity branches from Page');
      t.end();
    });

    t.test('🇵🇱::Extract entity Activities', async t => {
      const psDetail = fs.readFileSync(htmlFile);
      const expectedJSON = jsonfile.readFileSync('tests/data/pl/pi_003_activity.json');

      const output = await plScraper.extractEntityActivity(psDetail);

      t.deepEquals(output, expectedJSON, 'Extracted entity activities from Page');

      t.end();
    });

    t.end();
  });

  t.end();
});

test.test('Credit Services', async t => {
  test.test('POLAND:: Process CS Entity 001', async t => {
    const htmlFile = 'tests/data/pl/cs_001.html';
    t.test('🇵🇱::Extract entity details', async t => {
      const psDetail = fs.readFileSync(htmlFile);
      const expectedJSON = 'AXA UBEZPIECZENIA TUiR S.A.';

      const output = await plScraper.extractCSHeading(psDetail);

      t.deepEquals(output, expectedJSON, 'Extracted entity heading from Page');

      t.end();
    });

    t.test('🇵🇱::Extract entity body text', async t => {
      const psDetail = fs.readFileSync(htmlFile);
      const expectedJSON = 'Nazwa zestawienia: Zestawienie notyfikacji dotyczących działalności krajowych zakładów ubezpieczeń świadczące usługi w formie innej niż oddział w państwach UE/EEA';

      const output = await plScraper.extractCSBodyText(psDetail);

      t.deepEquals(output, expectedJSON, 'Extracted entity body text from Page');

      t.end();
    });

    t.test('🇵🇱::Extract entity table', async t => {
      const psDetail = fs.readFileSync(htmlFile);
      const expectedJSON = jsonfile.readFileSync('tests/data/pl/cs_001_table.json');

      const output = await plScraper.extractCSTable(psDetail);

      t.deepEquals(output, expectedJSON, 'Extracted entity table from Page');

      t.end();
    });

    t.end();
  });

  test.test('POLAND:: Process CS Entity 002', async t => {
    const htmlFile = 'tests/data/pl/cs_002.html';
    t.test('🇵🇱::Extract entity details', async t => {
      const psDetail = fs.readFileSync(htmlFile);
      const expectedJSON = 'HSBC Bank Polska S.A.';

      const output = await plScraper.extractCSHeading(psDetail);

      t.deepEquals(output, expectedJSON, 'Extracted entity heading from Page');

      t.end();
    });

    t.test('🇵🇱::Extract entity body text', async t => {
      const psDetail = fs.readFileSync(htmlFile);
      const expectedJSON = 'Nazwa zestawienia: Zestawienie notyfikacji dotyczących działalności transgranicznej banków krajowych na terytoriach państw należących do EOG';

      const output = await plScraper.extractCSBodyText(psDetail);

      t.deepEquals(output, expectedJSON, 'Extracted entity body text from Page');

      t.end();
    });

    t.test('🇵🇱::Extract entity table', async t => {
      const psDetail = fs.readFileSync(htmlFile);
      const expectedJSON = jsonfile.readFileSync('tests/data/pl/cs_002_table.json');

      const output = await plScraper.extractCSTable(psDetail);

      t.deepEquals(output, expectedJSON, 'Extracted entity table from Page');

      t.end();
    });

    t.end();
  });

  test.test('POLAND:: Process Art 70', async t => {
    const htmlFile = 'tests/data/pl/cs_art70.html';
    t.test('🇵🇱::Extract Art 70 details', async t => {
      const psDetail = fs.readFileSync(htmlFile);
      const expectedJSON = jsonfile.readFileSync('tests/data/pl/cs_art70.json');

      const output = await plScraper.reduceArt70(psDetail);

      t.deepEquals(output, expectedJSON, 'Extracted Art70 data from Page');

      t.end();
    });

    t.end();
  });

  t.end();
});