Bump weasyprint

[matmair]

This removes the pin as the issue that caused the pin seems to be resolved in django-waesyprint.
Might also close #4936, pending confirmation by @miggland

Closed #4936

[netlify]

✅ Deploy Preview for inventree-web-pui-preview canceled.

Name	Link
🔨 Latest commit	493877e
🔍 Latest deploy log	https://app.netlify.com/sites/inventree-web-pui-preview/deploys/655691ee09b1d80008e4ecc4

[miggland]

This issue was blocking using an upgraded django-weasyprint: fdemmer/django-weasyprint#71 . The fix was merged (fdemmer/django-weasyprint#70), but not released?

Either way: if Weasyprint can be successfully upgraded, #4936 can be closed.

[matmair]

You are right @miggland, we will have to push this

[SchrodingersGat]

The fix in django-weasyprint has actually been merged, but the release was not pushed to GitHub:

https://pypi.org/project/django-weasyprint/

[miggland]

Well spotted @SchrodingersGat - but as of 13 hours ago, that's been corrected as well it seems!

[SchrodingersGat]

@miggland yes I reached out to the maintainer of the project :)

[SchrodingersGat]

It appears that the rendered PDF does not contain the exact same information. If you are able to test that the generated PDF files are still good, we can adjust the single failing test

[miggland]

I have tried out (again) with the dev branch: SVGs don't show up.
With this branch, they look correct! All of the PDFs I've created in my test setup look fine. I used the standard demo data and made a few different labels/reports (also on sheets, or concatenated reports). The only thing I changed was to add an example with the SVG.

So: this seems to fix #4936, and looks good otherwise.

The test: I guess a more robust test would be to try and open the PDF with some tool (eg PikePDF), eg:
pdf = pikepdf.open(labelpath)
which gives exception pikepdf.PdfError if there is something wrong with PDF:

Alternative: pdftotext:

with open(labelpath, "rb") as f:
    pdf = pdftotext.PDF(f)
    filetext = " ".join(pdf)  # Join for multiple pages

But that will only work if there is actually text in the PDF..

Of course, both require installation of extra packages..

[matmair]

I am not sure adding a dependency is great but I also can not find a better solution, pretty much on the same page as @miggland.

I did not work with pikepdf - which is smaller than what I tried. I will adapt this PR to add a test based on it.

[matmair]

@miggland I added a co-author credit for you as the change is basically what you suggested above

[matmair]

The order of objects seems flaky between DBMS, a fix will follow shortly.

[matmair]

@SchrodingersGat this would be ready for review

[SchrodingersGat]

Thanks @matmair