Fix #495: resolve params in PDFStream.get_filters

[EvaSDK]

Pull request

Some PDF documents use reference to store filter params. Resolve them to allow proper extraction of content.

Below is a minimum code raising the problem with the document I was working with (different from #495) but it has PII so I cannot share it. I found #495 and reproduced the problem with it which is what the PR is now using but I kept this in the long description because that's what lead me to fixing the issue.

I switched building _filters to list-comprehension just for consistency with the code I added.

In [1]: import pdfplumber

In [2]: doc = pdfplumber.open("bill.pdf")

In [3]: s = doc.images[0]['stream']

In [4]: s.get_data()
---------------------------------------------------------------------------
TypeError                                 Traceback (most recent call last)
Cell In[4], line 1
----> 1 s.get_data()

File ~/.local/share/virtualenvs/pdfreader/lib/python3.11/site-packages/pdfminer/pdftypes.py:396, in PDFStream.get_data(self)
    394 def get_data(self) -> bytes:
    395     if self.data is None:
--> 396         self.decode()
    397         assert self.data is not None
    398     return self.data

File ~/.local/share/virtualenvs/pdfreader/lib/python3.11/site-packages/pdfminer/pdftypes.py:373, in PDFStream.decode(self)
    371     raise PDFNotImplementedError("Unsupported filter: %r" % f)
    372 # apply predictors
--> 373 if params and "Predictor" in params:
    374     pred = int_value(params["Predictor"])
    375     if pred == 1:
    376         # no predictor

TypeError: argument of type 'PDFObjRef' is not iterable

In [5]: s.get_filters()
Out[5]: [(/'FlateDecode', <PDFObjRef:21>)]

How Has This Been Tested?

Found a PDF exposing the problem in #495 and implemented unit test with it. I did not check if the PDF is freely reusable.

Checklist

• I have read CONTRIBUTING.md.
• I have added a concise human-readable description of the change to CHANGELOG.md.
• I have tested that this fix is effective or that this feature works.
• I have added docstrings to newly created methods and classes.
• I have updated the README.md and the readthedocs documentation. Or verified that this is not necessary.

[pavlokochkin]

worked for me, kudos @EvaSDK

[pietermarsman]

@EvaSDK Thanks for figuring this out! Your time and effort is appreciated.

@pavlokochkin Thanks for the confirmation.