BUG: Fix RGB FlateEncode Images(PNG) and transparency

[pubpub-zz]

Take the number of colors into account for PNG images

Properly process the mask for transparency:

Fixes #1787
Fixes #1599

Adds support for inline images extraction:

Fixes #1368
Fixes #1863

Additional changes:

• Process TIFF predictor 2
• Upgrades Pillow requirement to version 9.5 for Python 9.11

[codecov]

Codecov Report

❗ No coverage uploaded for pull request base (main@2eee5ed). Click here to learn what that means.
Patch coverage: 88.69% of modified lines in pull request are covered.

Additional details and impacted files

@@           Coverage Diff           @@
##             main    #1834   +/-   ##
=======================================
  Coverage        ?   93.79%           
=======================================
  Files           ?       34           
  Lines           ?     6919           
  Branches        ?     1364           
=======================================
  Hits            ?     6490           
  Misses          ?      280           
  Partials        ?      149           

Impacted Files	Coverage Δ
pypdf/filters.py	93.86% <87.73%> (ø)
pypdf/_page.py	92.83% <100.00%> (ø)
pypdf/_utils.py	100.00% <100.00%> (ø)
pypdf/constants.py	100.00% <100.00%> (ø)

☔ View full report in Codecov by Sentry.
📢 Do you have feedback about the report comment? Let us know in this issue.

[pubpub-zz]

test images ;
from resources/labeled-edges-center-image.pdf (PNG + transparency wirht reversed scale) (!! updated)

from pdf_font_garbled.pdf:
2nd page watermark (RGB+mask PNG):

Image45 part of resources in TPL2 in 2nd page (jp2 file => zipped to stored) (JPG + transparency => JP2)
p65.zip

[pubpub-zz]

tiff predictor 2:
from https://corpora.tika.apache.org/base/docs/govdocs1/977/977609.pdf
image :

[pubpub-zz]

images possible updates

[MartinThoma]

@pubpub-zz This PR introduces two changes: (1) The bugfix and (2) the new interface.

Could you please make a second PR just with the bugfix? I feel like that one can be merged today/tomorrow, but I want to have another look at the new interface.

[pubpub-zz]

@pubpub-zz This PR introduces two changes: (1) The bugfix and (2) the new interface.

Could you please make a second PR just with the bugfix? I feel like that one can be merged today/tomorrow, but I want to have another look at the new interface.

For tracking I've isolated the new images and 2 other PR following : inline_images and new function to replace images. However I'm not inclined to process only the bugs ones appart : I've worked all changes at the same time and I would not like to introduce regressions with a mod not being correctly reported.

[MartinThoma]

@pubpub-zz I messed something up when I fixed the merge conflicts 🙈

__________________________________ test_cmyk ___________________________________

    @pytest.mark.enable_socket()
    def test_cmyk():
        """Decode cmyk with transparency"""
        url = "https://corpora.tika.apache.org/base/docs/govdocs1/972/972174.pdf"
        name = "tika-972174.pdf"
        reader = PdfReader(BytesIO(get_pdf_from_url(url, name=name)))
        url_png = "https://user-images.githubusercontent.com/4083478/238288207-b77dd38c-34b4-4f4f-810a-bf9db7ca0414.png"
        name_png = "tika-972174_p0-im0.png"
        refimg = Image.open(
            BytesIO(get_pdf_from_url(url_png, name=name_png))
        )  # not a pdf but it works
        data = reader.pages[0].images[0]
        assert ".jp2" in data.name
>       assert list(data.image.getdata()) == list(refimg.getdata())
E       assert [(255, 255, 255, 0),\n (255, 255, 255, 0),\n (255, 255, 255,

[MartinThoma]

@pubpub-zz Can you please help me? I have no idea where to look 🙈

[pubpub-zz]

under analysis.

[pubpub-zz]

shoud be fixed. I had to upgrade pillow to 9.5 for python 3.11

[MartinThoma]

Thank you 🙏

I'm quickly checking if any of the other PRs is ready to merge and will make a release in less than an hour. This time we have awesome improvements for image handling thanks to your efforts. That was really good work 👍 🤗