BUG: PdfWriter.append fails when there are malformatted articles

CONTRIBUTORS.md

@@ -39,6 +39,7 @@ history and [GitHub's 'Contributors' feature](https://github.com/py-pdf/pypdf/gr
 * [Mérino, Antoine](https://github.com/Merinorus)
 * [Murphy, Kevin](https://github.com/kmurphy4)
 * [nalin-udhaar](https://github.com/nalin-udhaar)
+* [Noah-Houghton](https://github.com/Noah-Houghton) | [LinkedIn](https://www.linkedin.com/in/noah-h-554992a0/)
 * [Paramonov, Alexey](https://github.com/alexey-v-paramonov)
 * [Paternault, Louis](https://framagit.org/spalax)
 * [Perrensen, Olsen](https://github.com/olsonperrensen)

pypdf/_writer.py

@@ -3003,7 +3003,10 @@ def add_filtered_articles(
         for p in pages.values():
             pp = p.original_page
             for a in pp.get("/B", ()):
-                thr = a.get_object().get("/T")
+                a_obj = a.get_object()
+                if isinstance(a_obj, NullObject):
+                    continue
+                thr = a_obj.get("/T")
                 if thr is None:
                     continue
                 thr = thr.get_object()

tests/example_files.yaml

@@ -120,3 +120,5 @@
   url: https://github.com/user-attachments/files/21578875/layout-parser-paper-with-empty-pages.pdf
 - local_filename: issue-3429.pdf
   url: https://github.com/user-attachments/files/21711469/bomb.pdf
+- local_filename: issue-3508.pdf
+  url: https://github.com/user-attachments/files/23211824/repair-manual-thermo-230-300-350-2012-en.pdf

tests/test_merger.py

@@ -7,6 +7,7 @@
 
 import pypdf
 from pypdf import PdfReader, PdfWriter
+from pypdf._crypt_providers import crypt_provider
 from pypdf.generic import Destination, Fit
 
 from . import get_data_from_url
@@ -15,6 +16,10 @@
 PROJECT_ROOT = TESTS_ROOT.parent
 RESOURCE_ROOT = PROJECT_ROOT / "resources"
 
+USE_CRYPTOGRAPHY = crypt_provider[0] == "cryptography"
+USE_PYCRYPTODOME = crypt_provider[0] == "pycryptodome"
+HAS_AES = USE_CRYPTOGRAPHY or USE_PYCRYPTODOME
+
 sys.path.append(str(PROJECT_ROOT))
 
 
@@ -399,6 +404,14 @@ def test_articles_with_writer(caplog):
     assert r.threads[0].get_object()["/F"]["/P"] == r.pages[0]
 
 
+@pytest.mark.skipif(not HAS_AES, reason="No AES implementation")
+@pytest.mark.enable_socket
+def test_null_articles_with_writer():
+    data = get_data_from_url(name="issue-3508.pdf")
+    m = PdfWriter()
+    m.append(BytesIO(data))
+
+
 def test_get_reference():
     writer = PdfWriter(RESOURCE_ROOT / "crazyones.pdf")
     assert writer.get_reference(writer.pages[0]) == writer.pages[0].indirect_reference