PPF-1086: Invisible non-ascii text

[tieugene]

Version

PyPDFForm=3.0.0

Issue Description

I tried to fill form with non-ascii utf-8 text (cyrillic).
As a result text fields are invisible.
In combinations:

• adobe_mode=False/True
• flatten=False/True (with False text is visible when field is editing)
• replace registered fonts (Arial, Helveticaa) with Liberation-Sans
• register 'myfont', Liberation-Sans
• assign these fonts to all widgets
• using my own form

Empty.

PDF Form Template

pdf_samples/sample_template

sample.pdf

[chinapandaman]

Hey, thanks for posting.

I downloaded the sample.pdf you attached and couldn't seem to reproduce your issue. See screenshots:

Chrome:

Adobe Acrobat:

[tieugene]

Yes, Chrom and Firefox ok.
I tried PDF viewers - QPdfView, Okular, MuPDF.
There is flatten version in attach

sample_flatten.pdf

[chinapandaman]

Here are some test results for sample_flatten.pdf:

Adobe Acrobat:

Sejda:

VSCode's PDF extension:

They all seem to display your filled PDF fine.

I don't have the code snippet you used to generate it. Maybe try adobe_mode=False for the tools you listed?

[tieugene]

Acrobat Reader work finw for me too.
Sejda - not use.
VS code seem use chromium for rendering.

What not works:

• QPdfView (https://launchpad.net/qpdfview) - is in official repos of debian, fedora, epel (as minimum)
• Okular - in all Linux repos and Windows Magazine
• MuPDF - in all Linux repos

#!/usr/bin/env python3
import json
import sys
from PyPDFForm import PdfWrapper

ADOBE=False
FLAT=True

def main():
    if len(sys.argv) != 4:
        print(f"Usage: {sys.argv[0]} <pdf_form> <input_json> <out_file>")
        sys.exit()
    with (open(sys.argv[2], 'rt', encoding='utf-8') as in_file):
        data = json.load(in_file)
        PdfWrapper(sys.argv[1], adobe_mode=ADOBE).fill(data, flatten=FLAT).write(sys.argv[3])

if __name__ == '__main__':
    main()

input_json.json:

{
	"test": "Тест",
	"test_2": "Тест2",
	"test_3": "Test3"
}

[tieugene]

There is sample filled with qpdfview and screen

sample_qpdfview.pdf

Looks fine in all three desktop viewers above

[chinapandaman]

I think I know what the issue is.

The tools that you listed that weren't able to display the filled text all lack the ability to generate an appearance stream unless the PDF form is filled by them.

The appearance stream is stored under the /AP property of the field. For the PDF engine that Chromium uses, simply setting the /AP property to the value you fill the text field with will allow it to generate an appearance stream during rendering. For Adobe Acrobat, setting /NeedAppearances to True at the AcroForm level will tell it to re-generate appearance streams which is why the they can be rendered properly too.

However, I tried QPdfView and although if you fill a PDF form using it it will generate the right appearance stream, it doesn't seem to do any sort of regeneration of it if you open a PyPDFForm filled PDF form with it. This means I will have to somehow figure out a way of generating such appearance stream in the library code, which I'm not sure if I'm capable of. It's also weird that this issue seems to only happen when the text field is filled with non-ascii text, which is telling me that QPdfView somehow does generate an appearance stream...but for ascii text only?

Anyway, my bandwidth is quite occupied recently. But I will take a look at if I could generate appearance stream in the library code. But don't put too much expectation on this as I speculate this to be quite a challenging task, especially if I need to ensure that the appearance stream works with all the font style changes the library supports.

[tieugene]

Anyway, my bandwidth is quite occupied recently. But I will take a look at if I could generate appearance stream in the library code. But don't put too much expectation on this as I speculate this to be quite a challenging task, especially if I need to ensure that the appearance stream works with all the font style changes the library supports.

As an example - sample_template filled with Adobe Acrobat Reader.
Loooks ok in all of tools that I wrote.

sample_template_filled_AAR.pdf

[chinapandaman]

I understand that, and it's for the same reason I just explained: Adobe Acrobat will generate appearance streams when you fill PDF forms using it.

The difference between AAR and the tools you listed is that AAR will also generate appearance streams when opening PDF forms filled with PyPDFForm thanks to the /NeedAppearances property, but the tools you listed do not do that.

[tieugene]

Anyway, my bandwidth is quite occupied recently. But I will take a look at if I could generate appearance stream in the library code. But don't put too much expectation on this as I speculate this to be quite a challenging task, especially if I need to ensure that the appearance stream works with all the font style changes the library supports.

Maybe good way to replace field with read-only widgets?
Like:

• text field with text
• dropdown - text too
  etc

[chinapandaman]

Anyway, my bandwidth is quite occupied recently. But I will take a look at if I could generate appearance stream in the library code. But don't put too much expectation on this as I speculate this to be quite a challenging task, especially if I need to ensure that the appearance stream works with all the font style changes the library supports.

Maybe good way to replace field with read-only widgets? Like:

• text field with text
• dropdown - text too
  etc

What you described here was the old way of doing things before v3.0.0. You can read this post about it.

If you want to go back to the old watermark approach of filling, try ping your version back to v2.5.0.