When looking at #2389, I stumbled upon some oddities about how PDF permissions are handled/exposed (#2389 (comment)):

• While we have PdfReader.decode_permissions (introduced in ENH: Extract document permissions #320), we do not expose the corresponding permissions publicly, but only through PdfReader._encryption.values.
• It is a bit confusing to have both the PdfReader._encryption.values.Perms and the PdfReader._encryption.P attributes without any actual documentation.
  • Perms is the permissions dictionary with permission handlers. Our code at

    pypdf/pypdf/_encryption.py

    Line 1128 in 4423267

    values.Perms = encryption_entry.get("/Perms", ByteStringObject()).original_bytes

    and

    pypdf/pypdf/_encryption.py

    Line 776 in 4423267

    Perms: bytes

    considers this as bytes instead? (PDF 1.7 section 8.7.3 "Permissions")
  • P is the actual permission integer value. (PDF 1.7 section 3.5.2 "Standard Security Handler")

I tend to argue that providing a human-readable public API for the P permissions might indeed make sense. Relying on PdfReader._encryption might work, but being an internal API, this is not guaranteed to reliably work in the future as well.