Tesseract.js can't load language files on deployed server

[danisss9]

Describe the bug
Tesseract throws error when running in a deployed VM (ex: "https://vmdev-01:1000/DocScanner").

But works when running on localhost (ex: "https://localhost/DocScanner").

The code and deployment is the same. Both servers are running windows 10, IIS and ASP.NET Core 3.1 and for Frontend i'm using Angular 12.

To Reproduce
Steps to reproduce the behavior:

1. Setup Tesseract.js to run on browser (see code bellow)
2. Deploy WebApp on a VPS or Docker
3. Shows error loading languages

Expected behavior
Should work the same as in localhost.

Screenshots
Error Message:

"assets/scripts/tesseract" folder:

"assets/scripts/tesseract/lang-data" folder:

Package.json tesseract versions:

Desktop:

• OS: Windows 10
• Browser: Chrome
• Version: 102.0.5005.115 (Official Build) (64-bit)

Additional context

Tesseract.js setup code:

import { createWorker } from 'tesseract.js';

private async OcrImage(img: string) {
    const worker = createWorker({
      corePath:
        (this.baseUrl ?? '/') +
        'assets/scripts/tesseract/tesseract-core.wasm.js',
      workerPath:
        (this.baseUrl ?? '/') + 'assets/scripts/tesseract/worker.min.js',
      langPath:
        location.origin +
        (this.baseUrl ?? '/') +
        'assets/scripts/tesseract/lang-data',
      cacheMethod: 'none'
    });

    await worker.load();
    await worker.loadLanguage('eng');
    await worker.initialize('eng');
    await worker.setParameters({
      tessedit_char_whitelist: '0123456789',
    });

    const data = await worker.recognize(img);
    await worker.terminate();
    return data;
  }

[zerosdev]

I have this problem too, I have 2 VPS with Ubuntu 20.04, 1 server is working fine and the other sometimes get this error even though eng.traineddata already exists. I don't know what happened. Both servers have the same code

[Balearica]

@zerosdev As your issue only occurs "sometimes" it is probably related to the caching error discussed in #666. This should be fixed or significantly mitigated in version 4. In short, it looks like eng.traineddata can be corrupted, so when this happens you should either delete this file, or disable caching entirely (by setting cacheMethod to "none") although this may increase network use.

[314513535]

I have same issue. I guess this component can only work by running on node.js but not pure javascript.

[Balearica]

@314513535 It is entirely possible to load custom traineddata files using JavaScript in browser. If you are having trouble doing so and want support please provide a reproducible example of your issue in the form of a code snippet (that can be run without additional dependencies) or repository.

Here is an example of a program using custom training data in browser: https://github.com/scribeocr/scribeocr/blob/master/main.js#L758

[danisss9]

This seems to be fixed in version 4.