ROB: Improve handling for malformed cross-reference tables

pypdf/_reader.py

@@ -803,6 +803,7 @@ def _read_standard_xref_table(self, stream: StreamType) -> None:
                         )
                         generation = 65535
                         offset = -1
+                        entry_type_b = b"f"
                     else:
                         logger_warning(
                             f"entry {num} in Xref table invalid but object found",

tests/test_reader.py

@@ -1863,3 +1863,14 @@ def test_read_pdf15_xref_stream():
             match=r"^Trailer cannot be read: Limit reached while decompressing\. 1545392 bytes remaining\.$"
     ):
         PdfReader(BytesIO(data_modified))
+
+
+@pytest.mark.enable_socket
+def test_read_standard_xref_table__two_whitespace_characters_between_offset_and_generation():
+    """Tests for #3482"""
+    url = "https://github.com/user-attachments/files/22591813/helloworld.pdf"
+    name = "issue3482.pdf"
+
+    reader = PdfReader(BytesIO(get_data_from_url(url, name=name)))
+    assert len(reader.pages) == 1
+    assert reader.pages[0].extract_text() == "Hello World!"

tests/test_workflows.py

@@ -767,7 +767,7 @@ def test_image_extraction2(url, name):
             "tika-918137.pdf",
         ),
         (
-            "https://unglueit-files.s3.amazonaws.com/ebf/7552c42e9280b4476e59e77acc0bc812.pdf",
+            "https://github.com/user-attachments/files/22596566/7552c42e9280b4476e59e77acc0bc812.pdf",
             "7552c42e9280b4476e59e77acc0bc812.pdf",
         ),
     ],