PDF Document does not display greek characters in text field correctly when rendered in pdf.js

[gwtdevlpr]

Attached PDF file:

Configuration:

• Web browser and its version: Mozilla any version
• Operating system and its version: Any Os, Tested on macOS
• PDF.js version: v4.1.392 Latest
• Is a browser extension:

Steps to reproduce the problem:

1. Open the Attached PDF in https://mozilla.github.io/pdf.js/web/viewer.html
2. We don't see the greek text in textbox fields but some junk characters

What is the expected behavior? (add screenshot)
Display proper greek characters (Screenshot of same file opened in chrome)

What went wrong? (add screenshot)
Junk characters shown instead of proper greek characters in text fields

Link to a viewer (if hosted on a site other than mozilla.github.io/pdf.js or as Firefox/Chrome extension):

[choradt]

Attaching a sample document.
Copy.pdf

Working and non-working screenshot of the rendered document (Chrome and Firefox)

[calixteman]

When clicking on one of the fields, the value should be correct.
That said when rendering a fleld, we extract the text from its appearance to display it in the input but when it's focused then we use its value (from the V entry). Here the appearance is using a font and an identity-H encoding so it's why we extract a wrong string.

[gwtdevlpr]

On click of the field it shows correct value and on moving away, it goes back to junk characters again. Could it be fixed to show right value all the time?

[choradt]

When pdf.js is used to render pdf inside a viewer, fields are not clickable.

But even if they were, no-one could not expect a user, to click on 40 or 50 fields in a form just to be able to read them.

This must be solved in a way that characters display ok, without clicking inside the field.

[calixteman]

@Snuffleupagus, do you know what would be the exact criteria to guess that we won't able to guess the unicode string from the string in Tj and the font properties ?
I guess a missing ToUnicode doesn't help but there are likely some cases (for example with basic ascii strings) where it's possible to guess even without a ToUnicode.
My idea is to use this info when we're getting the text from the appearance:
https://github.com/mozilla/pdf.js/blob/master/src/core/annotation.js#L1234-L1243
to not use it.
In the pdf here, the font has a missing file, no ToUnicode and a cidEncoding set to Identity-H.