BUG: Make reader.get_fields also return dropdowns with options

[KourFrost]

Adding /Opt to the attributes list get_fields() will now add /Opt if it has it. For Combo Box /Ch it will list all of the available options in the drop down menu.

Closes #391

[MartinThoma]

I think I need to check where / how we use that dictionary to understand why you've added it. Could you point me to the place in code where adding the constant improves PyPDF2?

[KourFrost]

https://pypdf2.readthedocs.io/en/latest/_modules/PyPDF2/_reader.html#PdfReader.get_fields

reader.get_fields() uses that constants list to add information to the dictionary and only returns attributes from that list.
field_attributes = FieldDictionaryAttributes.attributes_dict()

By adding /Opt to that list its now included, and with my limited testing only shows up if the field has Opt attribute. It didn't break any of your other tests in pytest, so I figured that would be a good solution.

Another solution could be adding like an overloaded function to get_fields that takes an additional list of attributes, Then it wouldn't need to update the constant attributes list.

If one would want to see the items in a drop down/combo box you would need something like this:

for annot in reader.pages[0]["/Annots"]:
    obj = annot.get_object()
    if "/T" in obj.keys():
        if obj["/T"] == objectTitle:
            if "/Opt" in obj.keys():
                print(obj["/Opt"])

[MartinThoma]

Thank you for explaining it 🤗

Would you mind doing the following?

1. constants.py: Add a attributes and a attributes_dict method to CheckboxRadioButtonAttributes similar to FieldDictionaryAttributes. Yes, it would be simpler to add this single attribute directly to FieldDictionaryAttributes, but that would mean we mix two different things. The constants.py should be very close to various tables of the PDF specs.
2. constants.py: In the very bottom, there is a PDF_KEYS. Please add the CheckboxRadioButtonAttributes there. It ensures in the tests that there are no obvious typos.
3. _reader.py: Please change field_attributes = FieldDictionaryAttributes.attributes_dict() to field_attributes = FieldDictionaryAttributes.attributes_dict();field_attributes.update(CheckboxRadioButtonAttributes.attributes_dict())
4. _reader.py: Please change for attr in FieldDictionaryAttributes.attributes(): in a similar way so that we iterate over both
5. Apply black . to format the code. You might need to pip install black first.

I think that should do the trick while maintaining that constants.py represents various tables from the PDF specification. What do you think?

[KourFrost]

I like that as a solution. I'll get that implemented.

[codecov]

Codecov Report

Merging #1114 (6eeba04) into main (e1f9772) will increase coverage by 0.02%.
The diff coverage is 100.00%.

@@            Coverage Diff             @@
##             main    #1114      +/-   ##
==========================================
+ Coverage   91.94%   91.96%   +0.02%     
==========================================
  Files          24       24              
  Lines        4681     4694      +13     
  Branches      967      968       +1     
==========================================
+ Hits         4304     4317      +13     
  Misses        231      231              
  Partials      146      146              

Impacted Files	Coverage Δ
PyPDF2/_reader.py	90.92% <100.00%> (+0.03%)	⬆️
PyPDF2/constants.py	100.00% <100.00%> (ø)
PyPDF2/generic.py	91.74% <100.00%> (+0.01%)	⬆️

---

Continue to review full report at Codecov.

Legend - Click here to learn more
Δ = absolute <relative> (impact), ø = not affected, ? = missing data
Powered by Codecov. Last update e1f9772...6eeba04. Read the comment docs.

[KourFrost]

It didn't turn out to be as simple as we had hoped. Rewriting get_fields also required a similar change in _write_field and the field object it self. There might be some refactoring that we could do as get_fields loops through the same attributes as _write_field. _write_field is only called in _build_field. _build_field is only called by get_fields.

Also an issue with fields as a definition. Increasing the scope of what attributes define a field ended being the final trick. Fields objects define themselves off of the FieldDictionaryAttributes constants.

[MartinThoma]

Hey, just a short heads up: it's currently super hot in Germany. I have about 35°C in my apartment during the day and not below 28°C at night. This makes me pretty unconcentrated, meaning I can't look at as many things as I wished.

TL;DR : the review might take a few days.

[KourFrost]

its all good, stay safe man!

[MartinThoma]

Thank you for the contribution 🤗

[MartinThoma]

Very nice work @KourFrost ! I will make a release of PyPDF2 to PyPI today that contains your contribution!

If you want, I can also add you to the https://pypdf2.readthedocs.io/en/latest/meta/CONTRIBUTORS.html list - Just send me the line I should add for you :-)

[KourFrost]

Thank you! This is my first real contribution to an open source project, so I'm happy to help.

sure add me to the contributors list. just "KourFrost" linked to my github
https:/KourFrost