obdfcascrape/tests/scrape.cz.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 cheerio = require('cheerio');

const StaticServer = require('static-server');

const path = require('path');

const Czech = require('../ncas/cz');

const sourcePath = path.join(__dirname, 'data/cz/');

var server = new StaticServer({
  'rootPath': sourcePath, // required, the root of the server file tree
  'port': 7357, // required, the port to listen
  'name': 'my-http-server', // optional, will set "X-Powered-by" HTTP header
  // 'host': '10.0.0.100', // optional, defaults to any interface
  'cors': '*', // optional, defaults to undefined
  'followSymlink': true, // optional, defaults to a 404 error
  'templates': {
    'index': 'foo.html', // optional, defaults to 'index.html'
    'notFound': '404.html' // optional, defaults to undefined
  }
});

server.start(function () {
  console.log('Server listening to', server.port);
  console.log('Serving', server.rootPath);
});

test.skip('CZECH:: Captcha Handler', async t => {
  t.test('CZECH::Break the captcha', async t => {
    const expectedJSON = '095471'; // jsonfile.readFileSync('tests/data/se/ci001-detail.json');

    const czScraper = new Czech();

    await czScraper._initBrowser(true);

    czScraper.page = await czScraper.browser.newPage();

    await czScraper.page.goto('http://127.0.0.1:7357/captcha-001.html', { 'waitUntil': 'networkidle2' });

    // await czScraper.page.goto('https://apl.cnb.cz/apljerrsdad/JERRS.WEB45.LOGIN_A?p_lang=en&p_err=8', { 'waitUntil': 'networkidle0' });

    const result = await czScraper.captchaTest();

    await cz.browser.close();

    t.equals(result, expectedJSON, 'Busted captcha');
    t.end();
  });

  t.end();
});

test('CZECH:: Scrape a Payment Service', async t => {
  t.test('CZECH::Extract Basic Payment Services from Page', async t => {
    const psDetail = fs.readFileSync('tests/data/cz/ps001.html');
    const expectedJSON = jsonfile.readFileSync('tests/data/cz/ps001.json');

    const czScraper = new Czech();

    czScraper.urlPrefix = 'https://apl.cnb.cz/apljerrsdad/';

    const output = await czScraper.extractEntityDetails(psDetail);

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

    t.end();
  });

  t.test('CZECH::Extract More advanced details Payment Services from Page', async t => {
    const psDetail = fs.readFileSync('tests/data/cz/ps002.html');
    const expectedJSON = jsonfile.readFileSync('tests/data/cz/ps002.json');

    const czScraper = new Czech();

    czScraper.urlPrefix = 'https://apl.cnb.cz/apljerrsdad/';

    const output = await czScraper.extractEntityDetails(psDetail);

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

    t.end();
  });

  t.test('CZECH::Extract More Authorized activities for Payment Services from Page', async t => {
    const psDetail = fs.readFileSync('tests/data/cz/ps003_auth.html');
    const expectedJSON = jsonfile.readFileSync('tests/data/cz/ps003_auth.json');

    const czScraper = new Czech();

    czScraper.urlPrefix = 'https://apl.cnb.cz/apljerrsdad/';

    const output = await czScraper.extractEntityAuthority(psDetail);

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

    t.end();
  });

  t.end();
});

test('CZECH:: Scrape creditServices Authorised activitites', async t => {
  t.test('CZECH::Extract Authorized activities ( No 7 section, wih Matrix )', async t => {
    const psDetail = fs.readFileSync('tests/data/cz/cs_001.html');
    const expectedJSON = jsonfile.readFileSync('tests/data/cz/cs_001.json');

    const czScraper = new Czech();

    czScraper.urlPrefix = 'https://apl.cnb.cz/apljerrsdad/';

    const output = await czScraper.extractCreditServicesEntityAuthority(psDetail);

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

    t.end();
  });

  t.test('CZECH::Extract Authorized activities ( 7 Section no active, no Matrix )', async t => {
    const psDetail = fs.readFileSync('tests/data/cz/cs_002.html');
    const expectedJSON = jsonfile.readFileSync('tests/data/cz/cs_002.json');

    const czScraper = new Czech();

    czScraper.urlPrefix = 'https://apl.cnb.cz/apljerrsdad/';

    const output = await czScraper.extractCreditServicesEntityAuthority(psDetail);

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

    t.end();
  });

  t.test('CZECH::Extract Authorized activities ( With 7 Section and Matrix )', async t => {
    const psDetail = fs.readFileSync('tests/data/cz/cs_003.html');
    const expectedJSON = jsonfile.readFileSync('tests/data/cz/cs_003.json');

    const czScraper = new Czech();

    czScraper.urlPrefix = 'https://apl.cnb.cz/apljerrsdad/';

    const output = await czScraper.extractCreditServicesEntityAuthority(psDetail);

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

    t.end();
  });

  t.test('CZECH::Extract Authorized activities ( 4 tables including Matrix )', async t => {
    const psDetail = fs.readFileSync('tests/data/cz/cs_004.html');
    const expectedJSON = jsonfile.readFileSync('tests/data/cz/cs_004.json');

    const czScraper = new Czech();

    czScraper.urlPrefix = 'https://apl.cnb.cz/apljerrsdad/';

    const output = await czScraper.extractCreditServicesEntityAuthority(psDetail);

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

    t.end();
  });

  t.end();
});