ocr-space-api-wrapper/test/test.js

Node.js wrapper for ocr.space APIs.

const assert = require('assert');

const { ocrSpace } = require('../index');
const { base64string } = require('./base64string');

describe('Tests for OCR Space API Wrapper', () => {
  it('should throw if input is wrong', async () => {
    await assert.rejects(() => ocrSpace());
    await assert.rejects(() => ocrSpace(''));
    await assert.rejects(() => ocrSpace(1));
    await assert.rejects(() => ocrSpace(true));
  });
  it('should return results with a URL input', async () => {
    const res1 = await ocrSpace('http://dl.a9t9.com/ocrbenchmark/eng.png');
    assert.ok(res1.ParsedResults);
    assert.ok(res1.ParsedResults.length);
    assert.strictEqual(res1.OCRExitCode, 1);
    assert.strictEqual(res1.IsErroredOnProcessing, false);
    assert.strictEqual(res1.SearchablePDFURL, 'Searchable PDF not generated as it was not requested.');
  });
  it('should return results with a local file input', async () => {
    const res1 = await ocrSpace('./test/eng.png');
    assert.ok(res1.ParsedResults);
    assert.ok(res1.ParsedResults.length);
    assert.strictEqual(res1.OCRExitCode, 1);
    assert.strictEqual(res1.IsErroredOnProcessing, false);
    assert.strictEqual(res1.SearchablePDFURL, 'Searchable PDF not generated as it was not requested.');
  });
  // Seems like OCR Space API does not support base64 image input anymore
  it.skip('should return results with a base64 image input', async () => {
    const res1 = await ocrSpace(base64string);
    assert.ok(res1.ParsedResults);
    assert.ok(res1.ParsedResults.length);
    assert.strictEqual(res1.OCRExitCode, 1);
    assert.strictEqual(res1.IsErroredOnProcessing, false);
    assert.strictEqual(res1.SearchablePDFURL, 'Searchable PDF not generated as it was not requested.');
  });
  it('should return results with a local file input and options #1', async () => {
    const res1 = await ocrSpace('./test/eng.png', { apiKey: 'helloworld', language: 'eng' });
    assert.ok(res1.ParsedResults);
    assert.ok(res1.ParsedResults.length);
    assert.strictEqual(res1.OCRExitCode, 1);
    assert.strictEqual(res1.IsErroredOnProcessing, false);
    assert.strictEqual(res1.SearchablePDFURL, 'Searchable PDF not generated as it was not requested.');
  });
  it('should return results with a local file input and options #2', async () => {
    const res1 = await ocrSpace('./test/eng.pdf', {
      isCreateSearchablePdf: true,
      isSearchablePdfHideTextLayer: true,
      scale: true,
    });
    assert.ok(res1.ParsedResults);
    assert.ok(res1.ParsedResults.length);
    assert.strictEqual(res1.OCRExitCode, 1);
    assert.strictEqual(res1.IsErroredOnProcessing, false);
    assert.notStrictEqual(res1.SearchablePDFURL, 'Searchable PDF not generated as it was not requested.');
    assert.match(res1.SearchablePDFURL, /https?:\/\/.*\.pdf/);
  });
  it('should return results with a local file input and options with OCREngine 2', async () => {
    const res1 = await ocrSpace('./test/eng.pdf', {
      isCreateSearchablePdf: true,
      isSearchablePdfHideTextLayer: true,
      scale: true,
      OCREngine: 2,
    });
    assert.ok(res1.ParsedResults);
    assert.ok(res1.ParsedResults.length);
    assert.strictEqual(res1.OCRExitCode, 1);
    assert.strictEqual(res1.IsErroredOnProcessing, false);
    assert.notStrictEqual(res1.SearchablePDFURL, 'Searchable PDF not generated as it was not requested.');
    // @todo: this test returns res1.SearchablePDFURL="Not applicable | Generated by Python OCR V1.0"
    // assert.match(res1.SearchablePDFURL, /https?:\/\/.*\.pdf/);
  });
  it('should throw if the request is aborted', async () => {
    const controller = new AbortController();
    const { signal } = controller;
    const promise = ocrSpace('./test/eng.png', { signal });
    controller.abort();
    await assert.rejects(promise);
  });
});