Adding a test framework for sphinx-simplepdf

[kreuzberger]

First implementation of a tox based test framework, trying to handle issues #112 and #120 in current versions

[kreuzberger]

Currently there is no github action to perform tests, nor are lint tests etc working. The only tests working are e.g.
tox -e py313-sphinx82
which will lead to errors due to #112.

E       assert 16 == 0
E        +  where 16 = len(["ERROR: Content discarded: target points to undefined anchor ('url', ('internal', 'document-chapter1#section-1-1'))",...ion-1-1 for internal URI reference', 'ERROR: No anchor #document-chapter1#section-1-1 for internal URI reference', ...])

[kreuzberger]

for #120 and sphinx 7.4 the tests fails cause the links associcated with Chapter 1 should lead to page 4 and the Chapter 2 links to page 5. In the current version they links to page 3/4.

      text = extract_pdf_text(pdf_path)
    
>       assert """
    Table of Contents
    
    Contents:
    
    Chapter 1: Getting Started
    
    • Section 1.1
    • Section 1.2
    
    Chapter 2: Advanced Topics
    
    • Section 2.1
    • Section 2.2
    
    4
    
    4
    
    4
    
    5
    
    5
    
    5
    """ in text

[kreuzberger]

@ubmarco any suggestions?

[kreuzberger]

This is related to #119, also introducing a pyproject.toml file.

[ubmarco]

The PR #119 got merged, please rebase this

[kreuzberger]

This branch / pull request needs a review by a real python developer. I tried to merge by best knowledge, but due to my main scope (c++) i am not sure how to handle all the stuff in pyproject.toml.
E.g. the question if the dependencies for the documents should be an extra or an group dependeny and so on.

[kreuzberger]

The relations between the demo project and the doc projects seems to be covered now, but i have no idea thats the difference between the /Sphinx-SimplePDF-DEMO.pdf and the also referenced /Sphinx-SimplePDF.pdf

Some targets now work, e.g. make test with sphinx < 8.2, some are not (e.g. make linkcheck)

[kreuzberger]

❗ should not be merged until windows tests are also successfull. currently disabled but on work....

[kreuzberger]

Looking at the main branch, i see that weasyprint also does not work there.
The build jobs results are good, but if you look at the output, you see that weasyprint crashes with segfault.

WARNING: CalledProcessError in weasyprint, retrying
Command '['weasyprint', 'D:\\a\\sphinx-simplepdf\\sphinx-simplepdf\\docs\\_build\\simplepdf\\index.html', 'D:\\a\\sphinx-simplepdf\\sphinx-simplepdf\\docs\\_build\\simplepdf\\Sphinx-SimplePDF.pdf']' returned non-zero exit status 3221225477.

[kreuzberger]

Hi!
@ubmarco : I cannot properly finish this branch like i intended, cause the test do not work on windows. This is related to #130, the pdf builds for the doc project also do not build (status is green, but weasyprint return with nonzero exit code).
I do not know how to proceed, please lets discuss.

[kreuzberger]

Yeah, 😀 , this was hard work 💪 but glad to finish it at the end with success 🎉

@ubmarco , @danwos : This is just a beginning if you see our discussions at #83, but you see how important even for simple tests of the extension it is helpful.

PDF Generation out of the requirement documents is not just a gimmick, from our customers’ and managers’ perspective, this is a must-have acceptance criterion ❗

[aanagnosastronauticscom]

PDF Generation out of the requirement documents is not just a gimmick, from our customers’ and managers’ perspective, this is a must-have acceptance criterion ❗

Very true, internally, i have, a 3 digit number of tests covering PDF generation.
You simply can't give HTML to a great number of corporate and government entities.

[kreuzberger]

PDF Generation out of the requirement documents is not just a gimmick, from our customers’ and managers’ perspective, this is a must-have acceptance criterion ❗

Very true, internally, i have, a 3 digit number of tests covering PDF generation. You simply can't give HTML to a great number of corporate and government entities.

We do the same, currently we use libpdf for this in our test environment for the pdfs.
To avoid inter-extension dependencies (libpdf currently has mutually exclusive dependencies) i currently use just pdfminer.six here.

If you want it could be helpful to share your experience in #83 with tips and suggestions about what and how you test